Note: Descriptions are shown in the official language in which they were submitted.
-- 1 --
SPEECH SPEED CONVERSION ME;l~OD AND
SPI~:ECH SPEEI) CONVE:RSION APPARATUS
BACKGROUND OF THE INVENTION
The present invention relates to a speech
speed conversion method, a speech speed conversion
apparatus, and an electronic apparatus for modulating
the speed of a voice, and, particularly, to a technique
which is effective on application to a control technique
for using such an apparatus in conversation and so on.
As a device for aiding hearing sense for
persons hard of hearing, there have been, conven-
tionally, mainly used analog type hearing aids usinganalog circuits for processing the amplitude and
frequency characteristic of a voice. On the contrary,
research and development for making application of
digital signal processing to compensation for hearing-
impairment have been made eagerly in recent years. The
trend of the research and development has been described
in detail, for example, in "Application of Digital
Technique to Compensation for Hearing-Impairmentl~, ;
Journal of Acoustical Society of Japan (Vol. 47, No. 10, ;~
20~ pp.760-765, lg91), "Speech-perception aids for hearing-
impaired people: Current status and needed research",
J. Acoust. Soc. Am. (90(2), Pt. 1, Aug. 1991), and so
on.
To compensate for hearing loss, the amplifi-
cation of the amplitude of a speech signal and the
,.
;,
.
compression of a d~namic range are generally performed
with every frequency in accordance with the hearing
characteristic of a user. In ~he conventional analog
hearing aid, such a process is realized by an analog
circuit. On the other hand, in the digital hearing aid
developed in recent years, this process is realized by a
sof~ware such as a digital filter, or the like, so that
adaptation to the hearing characteristic of the user can
be made more in detail. .. ;
In the aforementioned trend, an attempt to
change only the speed of a voice by digital signal :~` -
processing without any change of the pitch of the voicè ` .`.
to thereby perform hearing aids of higher degree
covering the whole range of a hearing system inclusive
15 of the decline of language processing speed has been .`
made in recent years. Such a speech speed conversion .
technique has been described in detail, for example, in
"Development of Portable ~SP System for perorming ~ .
Speech Processing for the Aged", Technical Research
Report of Institute of Electronics, Information and
Communication Engineers of Japan (Vol. 92, No. 207 SP92-
54), "High--Quality Real-Time Speech Speed Conversion
System", ditto (SP92-55), and so on. .~;
Xn the aforementioned conventional techniques,
25 a broadcasted voice over the television/radio or the . .
like or a voice recorded in a tape recorder or the like .~
was used a~; the voice to be subjected to speech speed . `:
conversion. That is, the subject of speech speed .-~
''''~'; '"''.'' ''` ' ''. ,'' "' ~'," ''''`'``',., ''`",,'
-- 3 --
conversion was only a voice one-sidedly given to a
listener.
Taking into account the fact that the con-
ventional hearing aids can be used without distinction
of the input voice in kind, it is however preferable
that the speech speed conversion apparatus also can use
other voices than the aforementioned voices as the input
voice. Particularly, if the voice of a talker in
conversation can be heard slowly, the apparatus can be
used not only in the case of hearing perception aids for
aged or hearing-impaired people but also in the case of
hearing aids in conversation of a foreign language
unfamiliar to hearing-unhandicapped people, and so on.
An object of the present invention is provide
~15 a technique capable of reproducing a speech while
converting the speed of the speech at need.
Another object of the present invention is to
provide a technique in which raw data of a speech are
stored so that the speed of the speech can be converted
continuously on the basis of the raw data of the speech.
A further object of the present invention is
to provide in apparatus control unit for using a speech
speed conve:rsion apparatus in conversation or the like.
A further object of the present invention is
to provide i9 technique capable of widening the range of
application of the speech speed conversion apparatus.
A further object of the present invention is
to provide a technique capable of using a memory of the
.~
' ' ~ !
' ' ' : ' . : ~ ' '
. "; . ~ ' ' ' . , ' '` ~ , ' ; '
speech speed conversion apparatus effectively.
A further object o:E the present invention is
to provide a technique capable of returning a reading
pointer of the memory of the speech speed con~ersion
apparatus.
A further object of the present invention is
to provide a technique capable of controlling an AV ~-
apparatus connected to the speech speed conversion
apparatus.
A further object of the present invention is
to provide a technique capable of performing continuous
speech speed conversion in the speech speed conversion -
apparatus.
:
A further object of the present invention is
to provide a technique capable of attaining improvement
in handling property of the speech speed conversion -~
apparatus.
~ further object of the present invention is
to provide a technique capable of reducing electric
20 power consumed by the speech speed conversion apparatus. ~;
The foregoing and other objects and novel
features of the present invention will become clear from -
the description in this specification and the accompany~
ing drawings.
"'.:
SUMNARY OF THE INVENTION ::
Outlines of typical embodiments of the pressnt
invention disclosecl in this application will be
described in brief as follows.
(1) A speech speed conversion method for receiving an
input speech and changing the speed of the input speech
without any change of the pitch of the input speech
which comprises the steps of: carrying out a speech
speed conversion process for the input speech in a
period which is designated by a listener when speech
speed conversion is needed; and carrying out no speech
speed conversion process in the other period than the
designated period.
(2) A speech speed conversion apparatus which com-
prises: means for receiving an input a speech; a speech
speed conversion means for changing the speed of the
input speech; means for supplying an output of the
speech speed conversion means as a speech output to
listener's ears; a speech speed conversion switch
apparatus; and means adapted to output a speech while
changing the speech speed of the input speech in a
period in which the speech speed convQrsion switch is
turned on, and output an output speech without any
change of the speech speed of the input speech in the
other period in which the speech speed conversion switch
: ! iS turned off.
(3) A speech speed conversion method for encoding and
accumulating a raw speech, reading the accumulated
encoded speech and changing the speed of the speech
without any change of the pitch of the raw speech, which
comprises the steps of: carrying out a speech speed
'--
conversion process for an input speech in a period which
is designated when speech speed conversion is needed;
and carrying out no speech speed conversion process in
the other period than the designated period. ~;~
(4) A speech speed conversion apparatus which com-
prises: means for receiving a raw speech as an input
speech; a memory means for encoding and accumulating the
input speech; a speech speed conv~rsion means for `
reading the accumulated encoded-speech and for changing
the speed of the input speech; means for supplying an
output of the speech speed conversion means as a speech
output to listener's ears; a speech speed conversion
switch; and means adapted to output a speech while
changing the speech speed of the input speech in a : :
period in which the speech speed conversion switch is
turned on, and output an output speech without any
change of the speech speed of the input speech in the
other period in which the speech speed conversion switch ;:-
is turned off.
(5) In the speech speed conversion apparatus, the
memory means includes means for storing data by frame.
(6) The speech speed conversion apparatus ~urther
comprisès means for determining waveform expansion/ ~;:
reduction processes in the speech speed conversion .:
process on the basis of comparison between power of a
frame and a threshold provided as a variable.
(7) The speech speed conversion apparatus further
comprises a speech speed selection switch for s~lecting
. : - .. ~ .: : , . ; ,: . . .
: .. . .. -. .
the speed of the speech, and means for changing the
speed of the speech to the speech speed selected by the
speech speed selection switch.
(8) The speed conversion apparatus further comprises
means (AV control) for controlling an audio/video
apparatus.
(9) The speed conversion apparatus further comprises a
repeat switch, and means for repeating a reproduced
speech in a period in which the repeat switch is turned
cn.
(10) In the speech speed conver~ion apparatus, the
repeat means includes at least one of means for turning
back the speech by several seconds whenever the repeat
switch is pushed once, means for sometimes generating
intermittent sounds while the speech is turned back,
means for stopping the turning-back of the speech when
the speech reaches an end of a xing buffer, and means
for selecting the speech speed at the repeat time.
(11) In the speech speed convexsion apparatus, the
means for selecting the speech speed at the repeat time
has at least two modes of default speed value repeat,
slow repeat, fast repeat and gradually accelerated
repeat.
(12) The speech speed conversion apparatus further
comprises a catching-up means for adjusting the quantity
of a lag from the real time in a period in which the
stored information is reproduced in the case where the
lag from the real time is caused by a speech speed
conversion or repeat operation.
(13) In the speech speed co~version apparatus, the
catching-up means includes at least one of means for
starting catching-up when a slow reproduction mode is
terminated, means for starting catching-up when
reproduction is turned back to the point of time of the
start of a repeat after the repeat, means for selecting
the speech speed at the catching-up time, means for ~ `
automatically shifting the current mode to a through
mode for directly outputting the input speech when
catching-up is completed, and means for generating a
report signal sound (message) when catching-up is
completed.
(14) In the speech speed conversion apparatus, the
means for selecting the speech speed at the catching-up
time has at least one of means for making a nonstop skip
to the real time, means for catching up the real time
with fast hearing, and means for making a parallel
movement with a time lag.
(15) The speech speed conversion apparatus further
comp~ises at least one of the speech speed conversion
switch, a speech speed selection switch, a repeat
switch,~and a reset switch which are provided in a ;~
peripheral portion on a side surface of the speech speed
conversion apparatus so as to facilitate handling.
(16) In the speech speed conversion apparatus, the
reset switch includes means for stopping the repeating
or catching-up operation and making a skip to the real
time when the switch is turned on at the repeat or
catching-up time, and then shifting the current mode to
a through mode.
(17) In the speech speed conversion apparatus, the
speech speed conversion means is provided as a software
executed by a digital signal processor having an input
terminal for receiving an interruption request signal
from the outside, so that controlling of the speech
speed conversion process or switching of the rate of
speech speed conversion on the basis of the speech speed
convexsion switch is given to the digital signal
processor via the interruption request signal input
terminal.
(18) The speech speed conversion apparatus further
comprises ~eans for hearing the output speech through a
binaural headphone.
(19) A speech speed conversion apparatus which
comprises: a microphone for converting an acoustic
signal into an electric signal; an analog amplifier for
amplifying an output of the microphone; a low-pass
filter for removing high-frequency components from the ,~-
output of the analog amplifier; an A/D converter for
converting an output of the low-pass filter, which is an
analog signal, into a digital signal; a digital signal
processor for carrying out digital signal processing to
execute a speech speed changing process; a memory means
for holding input speech data and data obtained as a
result of signal processing; means for controlling the
-- 10 --
speech speed changing process executed by the digital
siignal processor; means for changing a processing
paramet~r; a D/A converter for converting digital speech
data into an analog value; a second low-pass filter for
removing high-frequency components from an output of the
D/A converter; a second analog amplifier for ampli.fying
an output of the second low-pass filter; and a headphone ..
for converting an output of the second analog amplifier
into an acoustic signal and supplying the acoustic
10 signal to both ears. ~:
(20) A speech speed conversion apparatus which
comprises: a microphone for converting an acoustic ~:.
signal into an electric signal; an analog amplifier for
amplifying an output of the microphone; a low-pass
filter for removing high-frequency components from an
output of the analog amplifier; an A/D converter for ..
converting an output of the low-pass filter, which is an .
analog signal, into a digital signal; a memory means for ... :::
holding input speech data and data obtained as a result
20 of signal processing; a digital signal processor for .
reading accumulated information and carrying out digital ;.
signal processing to execute a speech speed changing
process; means for controlling the speech speed changing .:
process executed by the digital signal processor; means .
2~ for changing a processing parameter; a D/A converter for
converting digital speech data into an analog value; a
second low pass filter for removing high-frequency -
components from an output of the D/A converter; a second
~.. -.. , . :, - . -, . , . ~ , . , . . . ~ , ., , , .. : - . , .
analog amplifier for amplifying an output of the second
low-pass filter; and a headphone for converting an
output of the second analog amplifier into an acoustic
signal and supplying the acoustic signal to both ears.
(21) In the speech speed conversion apparatus, the
speech speed conversion means carries out a series of
procedure over a whole frame repeatedly through a pipe-
line process by frame with use of a plurality of input
frame buffers, the series of procedure including:
applying a pitch extraction process to a leading portion
of the frame to detect the pitch of the leading portion;
transferring data of the length of one pitch thus
detected to output buffers; multiplying data of the
length of two pitches by a window function which changes
lS from 0 to 1 and by a window function which changes from
1 to 0; adding up respective data obtained by the
multiplications by the window functions to thereby
generate a reproduced wave pattern having a time length
of two pitches; inserting the reproduced wave pattern in
20 the rear of the preliminarily transferred data of the ~ :
length of one pitch; carrying out a pitch detection
process again while spearheaded by a position at a
distance of two pitches from the position preliminarily ~:-
subjected to the pitch extraction process to thereby :
perform pitch detection at the position; and transfer~
ring data of the length of n pitches (n is an integer)
based on the pitch length obtained by the final pitch~- :
detection to the output buffers.
. . ~ ~ :., .
'.''`''';
- 12 -
(22) A speech speed conversion apparatus wherein the
speech speed conversion means is executed only in the
case where average power of data in an input frame is
higher than a preliminarily set threshold, which data
contained in the frame are directly transferred to the
output buffers in the case where the average power is
lower than the threshold.
(23) In the speech speed conversion apparatus, a second
threshold is provided in the threshold process for the
a~erage power of data in the input frame so that ~hen a
frame having lower average power than the second
threshold is continued for a longer time than a pre-
liminarily set time threshold, data in the frame having
lower average power than the second threshold and
continued for a longer time than the time threshold are
forbidden to be transferred to the output buffers.
(24) In the speech speed conversion apparatus, the `
switch or each of the switches is constituted by a
switch which has a feeling of soft touch so that the
microphone does not pick up click noise of the switch.
(25) In the speech speed conversion apparatus, the
switches have respective surface forms different in
tactility so as to be identified without seeing.
(26) The speech speed conversion apparatus further
comprises a rustling prevention means for changing a
distance between the microphone and an apparatus body so
that the microphone does not touch clothes directly when
the apparatus body is put into a breast pocket in use.
i~:`,'" ' ' ,"., :, .': ' ' " ' ' . :'" ' ' ' ` ' '' ' , ~ : '`
(27) The speech speed conve:rsion apparatus further
comprises a display means which is provided at a
predetermined position of the speech speed conversion
apparatus so that a quantity of a time lag from the real
time can ke indicated visually.
(28) In the speech speed conversion apparatus, a ring
buffer is used as the memory means, and the apparatus
further comprises means for managing a lag time by a
counter indicating a ~ime lag on the ring buffer.
(29) In the speech speed conversion apparatus, a
standby mode for lowering the clock cycle of the
processor and carrying out the same process as in the
through mode is provided besides the through mode.
(30) The speech speed conversion apparatus further
comprises an electric source switch operated at three
stages consisting of an ON stage, an OFF stage and an
ON-OFF intermediate stage, and an electric source supply
means operated in an analog through mode in which analog
input-output systems are short-circuited so as to be
directly connected to each other to thereby stop elec-
tric source supply to a digital processing system
:
between the analog input-output systems when the switch ~ ~
:: :: ~ - .: ~, .
is adjusted to the intermediate stage. ~ ~-
(31) A telephone in which the speech speed conversion
25 means as defined in any one of the above paragraphs (2), ;
(4) and (3Q~ is provided between a handset of the
telephone and a body of the telephone. i;~
(32) A telephone line switching system in which the
;..
`,' ~
- 14 -
speech speed conversion means as defined in any one of
the above paragraphs (2), (4) and (30) is provided in a
telephone line switching system.
According to the feature described in the
5 above paragraphs (1) and (2), in the case where a speech ~:
is inputted and the speed of the speech is changed ::.
without any change of the pitch of the input speech, a
speech speed conversion process is carried out for the
input speech in a period which is designated when speech
speed conversion is needed, but no speech speed conver-
sion is carried out in the other period. Accordingly,
the speech speed conversion apparatus can be used not
only for a voice such as a radio voice one-sidedly given
to a listener but also in the situation of conversation, ~
15 so that a voice to be subjected to speech speed `.
conversion can be selected by the listener without any
disturbance of listener's own speech.
Further, in a hearing aid, a foreign language
learning machine, a telephone, or the like, talker's
voice can be heard at a slow speech speed without any
change of the characteristic of the talker's voice.
According to the feature described in the
above paragraphs (3) and (4), in the case where a raw
speech encoded and accumulated is read to change the
speed of the speech without any change of the pitch of
the raw speech, or the like, a speech speed conversion
process is carried out for the input speech in a period
which is designated when speech speed conversion is
- 15 -
needed, but no speech speed conversion is carried out ln
the other periodO According:Ly, in addition to the
effect provided by the paragxaphs (1) and (2)l there can
be provided effective use of the memory, a raw speech
repeat function, a voice memory f~nction, a repeat
speech speed conversion function, a fast hearing
reproduction function, and so on.
According to the feature described in the
above paragraph (5), data are stored by frame, so that
writing/reading efficiency can be improved.
According to the feature described in the ;~
a~ove paragraph (6), the determination of waveform
expansion/reduction process, silent-part elimination
process, etc. in the speech speed conversion process is
performed based on comparison between power of a frame
and a threshold, and the threshold is changed in accord- i`~
ance with the loudness of the input speech. According- ;;;
ly, the speech speed conversion process can be carried
out in accordance with the environmental condition in
use.
According to the feature described in the
above paragraph (7), in the speech speed conversion
apparatùs, there are provided a speech speed selection -~-
switch for selecting the speed of the speech, and means
for changing the speed of the speech to the speech speed
selected by the speech speed selection switch. Accord~
inglyl the speed of the speech to be heard can be ;
selected by the li~tener's own will.
"~ ~ ,
''.,-' ,'"" ~.
- 16 -
According to the feature described in the
above paragraph (8), means (AV control) for controlling
an audio/video apparatus is provided in the speech speed
conversion apparatus. Accordingly, a series of opera-
tion in which a signal for pausing the reproducingoperation of the external apparatus is issued to tempo-
rarily stop the inputting of the speech to the speech
speed conversion apparatus when the memory capacity is
insufficient and in which the outputting of the pause
s~gnal is stopped to re-start the inputting of the
speech from the external apparatus when there is some
free area in the memory, is repeated irrespective of the
expansion/reduction ra~e in the speech speed conversion.
As a result, use of speech speed conversion can be
continued for a long time.
~ ccording to the feature described in the
above paragraphs (9) to (11), in the speech speed
conversion apparatus, there are provided a repeat switch
and means for repeating a reproduced speech in a period
in which the repeat switch is turned on. Accordingly,
the speech speed conversion of the repeat speech can be
carried out.
According to the feature described in the
above paragraphs (12) to (14), a catching-up means for
catching up the speech to a position of stored infor-
mation to be heard is provided in the speech speed
conversion apparatus. Accordingly, widening of the
range of application of the speech speed conversion
apparatus, reduction in operating time, improvement in
handling property, and so on, can be a~tained.
According to the feature described in the
above paragraphs (15) and (16), at least one of the
speech speed conversion switc:h, speech speed selection
switch, repeat switch and reset switch is provided in a
peripheral portion on a side surface of the speech speed
conversion apparatus so as to perform handling easily.
Accordingly, widening of the range of application of the ~ ~
10 speech speed conversion apparatus, reduction in operat- -
ing time, improvement in handling property, and so on, -
can be attained.
According to the feature described in the
above paragraphs (17) to (23), the speech speed conver-
sion means is provided as a software executed by a
digital signal processor having an input terminal for ~
receiving an interruption request signal from the ~ ;
outside, so that controlling of the speech speed ~ ;
:
conversion process or switching of the speech speed
conversion rate on the basis of the speech speedconversion switch is given to the digital signal
processor via the interruption request signal input
terminal.
Further, the aforementioned speech speed
25 conversion means carries out a series of procedure over -
a whole frame repeatedly through a pipe-line process by
frame with use of a plurality of input frame buffers,
the series of procedure including: applying a pitch ;
" ~
- 18 -
extraction process to a leading portion of the frame to
detect the pitch of the lead.ing portion; transferring
data of the length of one pitch thus detected to output
buffers; multiplying data of the length of two pitches
by a window function which changes from 0 to 1 and by a
window function which change,s from 1 to 0; adding up
respective data obtained by the multiplications by the
window functions to thereby generate a reproduced wave
pattern having a time length of two pitches; inserting
the reproduced wave pattern in the rear of the pre-
liminarily transferred data of the length of one pitch;
carrying out a pitch detection process again while
spearheaded by a position at a distance of two pitches
from the position preliminarily subjected to the pitch
extraction process to thereby perfoxm pitch detection at
the position; and transferring data of the length of n
pitches (n is an integer) based on the pitch length
obtained by the final pitch detection to the output
buffers.
Further, the speech speed conversion means is
executed only in the case where average power of data
in an input frame is higher than a preliminarily set
threshold, which data contained in the frame are
directly transferred to the output buffers in the case
where the average power is lower than the threshold.
Further, a second threshold is provided in the
threshold process for the average power of data in the
input frame so that when a frame having lower average
~ .
-- 19 --
power than the second threshold is continued for a
longer time than a preliminaxily set time threshold,
data in the frame having lower average power than ~he
second threshold and continued for a longer time than
the time threshold are forbidden to be transferred to
the output buffers. .
By the aforementioned configuration of the
speech speed conversion means, there can be attained
improvement in speech speed conversion efficiency and
prevention of lowering of the quality of the reproduced
speech.
According to the feature described in the
above paragraph (24), the microphone does not pick up
click noise of each switch, so that loud noise at the
.: . .
15 time of the manipulation of the switch can be prevented. -
According to the feature described in the ;~
. .
above paragraph (25), the switches have respective
surface formed diferent in tactility so as to be
identified without seeing, 50 that handling property can ~ ~
20 be improved. - `
According to the feature described in the ` i
above paragraph ~26), there is provided means for
preventing the rustle of clothes in contact with the
microphone, so that entrance of noise can be reduced.
According to the feature described in the
above parayraph (27), a display means is provided at a
predetermined position of the speech speed conversion
apparatus so that the quantity of a time lag from the;-~
", .:
., ~,
- 20 -
real time can be indicated visually. Accordingly,
reduction in operating time, improvement in handling
property, and so on, can be attained.
According to the feature described in the
above paragraph (28), a ring buffer is used as the
memory means, and there is provided means for managing a
lag time by a counter indicating a time lag on the ring
buffer. Accordingly, the repeat process, the catching-
up process, and so on, can be carried out easily.
According to the feature described in the
above paragraph (29), a standby mode is provided besides
the through mode, so that reduction in consumed electric
power can be attained.
According to the feature described in the
above paragraph (30), there is provided an electric
source switch operated in three stages consisting of an
ON stage, an OFF stage and an ON-OFF intermediate stage
so that an analog through mode is provided. Accord-
ingly, reduction in electric power can be attained.
According to the feature described in the
above paragraph (31), the speech speed conversion means
is provided between a handset of a telephone and a body
of the telephone. Accordingly, a speech to be subjected
to speech speed conversion can be selected by the
listener without any disturbance of the listener's own
speech.
Further, in the telephone, the voice can be
heard at a slow speech speed without any change of the
... ;.. , . .. ~ .. . , .. ~ . - ,.
.: . .. ~ i .. :, . . .. . . . . .
- 21 -
characteristic of the talker's voice.
According to the feature described in the
above paragraph ~32), the speech speed conversion means
is provided in a telephone l:inP switching system.
Accordingly, the voice to be subjected to speech speed
conversion can be selected by the listener without any
disturbance of the listener's own speech.
Sill further advantages of the present
in~ention will become apparent to those of ordinary
sliill in the art upon reading and understanding the
following detailed description of the preferred and
alternate embodiments.
'` ',:'~'
BRIEF DESCRIPTION OF THE DRAWINGS
; :~',',
Fig. 1 is a block diagram showing the
schematic structure of internal circuits according to
present invention;
Fig. 2 is a graph for explaining a speech
speed conversion process executed within a DSP according
to present invention;
Fig. 3 is a graph for explaining the concept
of a threshold process according to present invention;
Fig. 4 is a view showing the form of use of
the speech speed conversion apparatus according to
present invention;
Fig. 5 is a flow chart showing the control
procedure of the speech speed conversion apparatus ,,;
according to present invention;
, , . -
,;,', .
- 22 -
Fig. 6 is a front plan viewed from the front
of the speech speed conversion apparatus according to
present invention;
Fig. 7 is a back plan viewed from the back of
the speech speed conversion apparatus according to
present invention;
Fig. 8 is a top plan viewed from the top of
the speech speed conversion apparatus according to
present invention;
Fig. 9 is a left plan viewed from the left of
the speech speed conversion apparatus according to
present invention;
Fig. lO i~ a right plan viewed from the right
of the speech speed conversion apparatus according to
present invention;
Fig. ll is a block diagram showing the
functional structure of the speech speed conversion
apparatus according to present invention;
Figs. 12A and 12B are typical graphs for
explaining a compression process in a speech compression
portion according to present invention;
Fig. 13 is a flow chart showing the procedure
of a main process according to present invention;
Fig. 14 is a flow chart to be continued from
the flow chart of Fig. 13;
Fig. 15 is a state transition view typically
showing transition between respective modes according to
present invention;
.:.. , . ~ . , . . . . :
~:; . .. .
Fig. 16 is a flow chart showing the procedure
of a reading pointer return routine according to present
invention;
Fig. 17 is a flow chart showing the procedure
of a one-frame waveform expansion/reduction process
according to present invention;
Fig. 18 is a flow chart to be continued from
the flow chart of Fig. 17; -
Fig. 19 is a flow chart showing the procedure -
10 OL a parameter setting process according to present ~ ;
invention;
Fig. 20 is a view for explaining the data
compression process according to present invention;
Fig. 21 is a view for explainin~ the data
15 compression process according to present invention; ~;
Fig. 22 is a view for explaining the data
compression process according to present invention;
Yig. 23 is a flow chart showing the procedure -~
of the total operation of the speech speed conversion
apparatus provided with a continuous speech speed
conversion means according to present invention; , ;
Fig. 24 is a flow chart to be continued from -
the flow chart of Fig. 23;
E~ig. 25 is a flow chart showing the procedure "`;'~
of the total operation of the speech speed conversion
apparatus provided with a continuous speech speed
conversion means according to present invention;
Fig. 26 is a flow chart to be continued fxom
,.. .~ .. ~ :
" "
- 24 -
the flow chart of Fig. 25;
Fig. ~7 is a typical view for explaining an
accelerator type switch used in the continuous speech
speed conversion means according to present invention;
Fig. 28 is a block diagram showing the
functional structure of the speech speed conversion
apparatus provided with an AV control means according to
present invention;
Fig. 29 is a view for explaining the operation
of the AV control means according to present invention;
Fig. 30 is a flow chart showing the procedure
of a main process in the speech speed conversion
apparatus provided with the AV control means according
to present invention;
Fig. 31 is a flow chart to be continued from
the flow chart of Fig. 30;
Figs. 32A to 32C are views for explaininy the
arrangement of the microphone in the speech speed
conversion apparatus according to present invention;
Fig. 33 is a view showing the structure of a
modified example according to present invention;
Fig. 34 is a view for explaining a lag time
display means in thé speech speed conversion apparatus
according to present invention;
Fig. 35 is a diagram for explaining an
electric source device in the speech speed conversion
apparatus according to present invention;
Fig. 36 is a diagram for explaining an
- 25 -
embodiment in which the speech speed conversion means
according to the present invention is applied to a
telephone; and
Fig. 37 is a diagram for explaining an
embodiment in which the speech sp~ed conversion means
according to the present invention is applied to a
premises broadcasting system.
~ .', ~',
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Embodiments of the present invention will be
::
described below in detail with reference to the
drawings. ;
In the all drawings for explaining the
embodiments, parts having the same function are marked
by the same reference character to omit the repeated -
description thereof.
Fig. 1 is a block diagram showing the
schematic structure o internal circuits according to
the present invention. The reference numeral 1
designates a DSP (Digital Signal Processor); 11, a
software for performing a speech speed conversion
process; 12, a serial port; 13, a terminal for external
interruption flag; 14, a flag register; 2, a memory
(output buffer); 3, a selector switch; 4, a PTL ~Push-
To-Listen) switch; 5, an A/D con~erter; 6, a D/A con-
verter; 7, a low-pass filter; 8, a low-pass filter; 9,
an analog amplifier; 10, an analog amplifier; 321, a
microphone; and 325, a binaural headphone ~earphone).
In a speech speed conversion apparatus
according to this embodiment" as shown in ~ig. 1, a
voice is inputted to the microphone 321 and outputted as
a voice signal (an electric isignal). This voice signal
is inputted via the amplifier 10 and the low-pass filter
7 to the A/D converter 5, in which the voice signal is
converted from an analog value into a digital value at
intervals of a time set in advance.
The voice signal convertied into a digital
value as described above is inputted to the DSP 1.
Then, the speech speed conversion process of the yoice
signal is realized by the software 11 on the DSP 1. The
PTL switch 4 is connected to the external interruption
flag terminal 13 contained in the DSP 1, so that the
state of the PTL switch 4 is expressed as a numerical
value of the flag register 14 which is provided in the
inside of the DSP 1 so as to correspond to this terminal
13. In the software 11 on the DSP 1, a judgment in
accordance with the numerical value of the flag register
14 is made as to whether the speech speed conversion
process is to be performed or not to be performed.
The digital voice data subjected to the speech
speed conversion process is stored in the output buffer
memory 2. The D/A converter 6 converts the data of the
2~ output buffer memory 2 from a digital value into an
analog value at intervals of a time set in advance. The
analog signal obtained by this conversion is inputted
via the low-pass filter 8 to the analog amplifier 9 and
- 27 -
outputted as a voice from the binaural headphone 325 in
lîstener~s favorite amplitude of speech signal.
In this embodiment, two kinds of switches are
prepared for the PTL switch 4. One thereof is a switch
in which current conduction is made as long as a
pushbutton is pushed. The other is a switch in which
th~ current conduction state is maintained though the
hold of the pushbutton is released. The former is used
in the case of conversation whereas the latter is used
in the case of continuous speech speed conversion of a
one-sidedly given voice such as a radio voice which is
conventional utilization, and the like. Further in this ;;
embodiment, the selector switch 3 as well as the PTL
switch 4 is connected to the external interruption flag ;
.
terminal 13 contained in the DSP 1. The numerical value
of the flag register 14 is changed by the changeover of
the selector switch 3, so that the software 11 changes ;~
the expansion rate of the speech speed conversion
process in accordance with this numerical value.
Fig. 2 is a view for explaining the speech
speed conversion process which is performed in the DSP 1
in this embodiment. The speech speed conversion process
in this embodiment is a method of detecting the pitch ~;~
(basic per:iod) of a voice signal and expanding the
length of a waveform with the detected pitch as a unit,
in which a voice data set of the order of tens of
milliseconds (hereinafter referred to as a frame) is
made a unit for one process. Accordingly, at least two
I .
- 28 -
frame length input buffers are pxepared in the inside of
the DSP 1 so that while data from the ~/D converter is
inputted to one buffer, data stored in the other buffer
is processed (pipe-line process). After processed, data
5 i5 stored in the output buffer 2 having a sufiificiently
large capacity. The procedure of processing data in
each fram~ is as follows.
First of all, (a) a pitch extracting process
(not shown) is applied to the head portion of the frame
to thereby detect the pitch of this portion.
(b) Then, the thus detected data of the length of
one pitch is transferred to the output buffer 2.
(c) Then, data of the length of two pitches is
multiplied by a window function which changes from O to
1 and by a window function which changes from 1 to 0.
The on-data positions from which the multipli
cations by the window functions are started are however
shifted by one pitch. Then, data as respective results
of the multiplications by the window functions are added
to each other to generate a reproduced wave pattern
having the time length of two pitches and put the pat-
tern in the rear of the one-pitch data which was trans-
ferred in advance.
(d) Then, a pitch detecting process (not shown) is
carried out again in the condition in which a position
two pitches away from the on-data position previously
subjected to the pitch extracting process is at the
head, so that pitch detection in this position is
", , . . , , ,. ,: , , . , ., .. - . . - . : - -
- 29 -
performed. Because the voice pitch generally always
varies, a pitch different from the previously detected -~
pitch is obtained in the second detection.
(e) Data of the length of n pitches is transferred
to the output buffer with the pitch length obtained by
this final pitch detection as a unit.
The aforementioned procedure of from (a) to
(e) is repeated over the whole frame.
Because the pitch length depends on the input
voice, the number of repeats in one frame is not con-
stant. Further, different expansion rates are realized
by changing the value of n in the aforementioned step
(e). For example, data of the length of four pitches is
generated from data of the length of three pitches in
the input buffers in the condition of n=1, so that the
expansion rate becomes 4/3=1.33 times. Similarly, the
expansion rate becomes 1.50 times in the condition of
n=0 and 1.25 times in the condition of n=2.
Further, in this embodiment, the afore- -~
mentioned speech speed conversion process in Fig. 2 is
not always applied to all frames but the aforementioned
process in Fig. 2 is applied only in the case where the
calculated average power of each frame exceeds a thresh-
old Th which was set in advance. Data in a frame having
power not exceeding the threshold Th is therefore trans-
ferred to the output buffer in its original condition.
FigO 3 shows a concept of this threshold process.
In Fig. 3, the portion in which the power of
- 30 -
each frame exceeds the threshold Th is expressed as a
duration of expansion. Because the leading and trailing
portions of the voice signal are not processed but
outputted in their original condition by this threshold
S process, there is an advantage in that voice character-
istic contained in the leading and trailing of the
voice, for example, consonantal characteristic, is not
destroyed.
Further, in this embodiment, a second thresh-
old To is provided in the threshold process for theaverage power of each frame as shown in Fig. 3. In the
case where a Erame having power lower than this second
threshold To is continued for a time not smaller than
one second, the frame having power lower than the
threshold To continuously for a time not smaller than
one second is therefore processed so as not to be
outputted. Accordingly, reduction in the quantity of
data stored in the output buffer is attained.
In Fig. 3, this not-outputted portion is ~
20 expressed as a duration of elimination. ~t the output `
buffer 2, data are one by one outputted to the D/A
converter 6 at regular time intervals in parallel with
the writing oE the speech-speed-conversion-processed
data at once each frame. Addresses in the output buffer
25 2 are set in the form of a ring so that the last address ~;
is continued to the first address. -;~ ;
Accordingly, in this ring-like address space,
an operation is carried out so that an address pointer
:.,"
- 31 -
Po which points data to be fed to the D/A converter runs
after an address pointer Pi which points the destination
of the writing of the speech-speed-conversion-processed
data. In this embodiment, P:i will overtake Po sooner or
later because the speed of P:i is higher than the speed
of Po. At this point of time, information which has
been stored in the output bu:Efer 2 is not outputted but
rewritten.
Accordingly, the time from the start of the
speech speed conversion operation to this state becomes
a time length of the input voice which can be tackled by
the speech speed conversion process of this embodiment.
The reduction in the quantity of data based on the
aforementioned threshold To has an effect that this time
length which can be tackled is made long.
Further, the signal processing method in the
speech speed conversion process explained above with
reference to Figs. 2 and 3 has been reported in
"Evaluation of Speech Speed Conversion Method by
Hearing-Impaired People" Technical Research Report of
Institute of Electronics, Information and Communication
Engineers of Japan, SP92-150 (1993-03) or "Discussion of
Speech Speed Conversion Method using Portable DSP
System" Proceedings of Acoustical Society of Japan
(March 1993), 1-7-6.
Fig. 4 is a view showing the form of use of
the speech speed conversion appaxatus according to this
embodiment. ~lthough Fig. 4 shows the case where the
PTL switch 4 is disposed on the upper surface of ~he
apparatus, it is a matter of course that the position of
the arrangement thereof may be replaced by another
position. On the other hand, the selector switch 3 for
changing the expansion rate of speech speed conversion
is prepared in a side of the PTL switch 4. Because the
selector switch 3 as well as the PTL switch 4 is
provided so that the state of the selector switch 3 can
be observed through the external interruption flag
terminal of the DSP 1 from the software on the DSP 1,
the value of n in the aforementioned speech speed
conversion process is changed in accordance with the
state of the selector switch 3 when the PTL switch 4 is
pushed. The expansion rate can be changed for every
speech by opexating the PTL switch 4 and the selector
switch 3 alternately.
Fig. 5 shows the aforementioned control
procedure expressed in a flow chart. In the speech
speed conversion process, for example, a flame of the
time len~th of the order of tens of milliseconds is used
as a unit of processing. The A/D conversion and D/A ;~
conversion are processes which are carried out at
regular in~ervals of a smaller time pitch than that, for
example, at regular intervals of a time pitch of the
25 order of, Eor example, tens of microseconds. As shown -
in Fig. 5, the A/D conversion, the D/A conversion and;~
their attendant process are realized as an interruption
process. While the speech speed conversion process and ~ ;
,, . ,, ,~
a process of waiting for interruption are carried out,
the interruption process is carried out in accordance
with an in~erruption signal from the serial port to
which the ~/D converter and the D/A converter are
connected.
As is obvious from the above description, in
accordance with this embodiment, the speech speed
conversion apparatus can be used not only for the voice
one-sidedly given to a listener like a radio broadcast-
ing voice but also i~l the situation of conversation, sothat the listener can select the voice subjected to the
speech speed conversion without any disturbance of
listener's own speech.
Further, the speech speed conversion apparatus
of this embodiment can be used to compensate for the
deterioration of voice hearing ability as observed in
the aged or the like. It is further needless to say
that the apparatus can be used even in the situation in
which a listener who has no difficulty in hearing hears
an unfamiliar foreign language.
Referring to Figs. 6 through lO, the external
appearance structure of the speech speed conversion
apparatus according to the present invention is shown in
the following. Fig. 6 is a front plan view from the
2S front; Fig. 7 is a back plan view from the back; Fig. 8
is a top plan view from the top; Fig. 9 is a left plan
view from the left; and Fig. lO is a right plan view
from the right.
. ~.. ' .,,.. .~, . .. .,, ,, . ' . . , , ., , ,.. , ., -... .. . .
- 34 -
In Figs. 6 to 10, the reference numer~l 101
designates a body of the speech speed conversion
appara~us; 102, a back cover; 103, a finger stop hollow;
104, a slow switch (slow pushbutton); 105, a repeat
switch (repeat pushbutton); 106, a reset switch (reset
pushbutton); 321, a microphone; 108, a voice volume;
109, an electric source switch; 110, an earphone
terminal; 111, an external input terminal; 112, an AV
control terminal; and 113, a speech speed changeover
switch (speech speed setting switch).
As shown in Figs. 6 to 10, in the speech speed
conversion apparatus of this embodiment, the slow switch :~
104, the repeat switch 105 and the reset switch 106 are ~-
provided in positions where the body 101 of the speech
speed conversion apparatus is easy to operate with one
hand, for example, in the front upper side portion, and .~` ~
the speech speed changeover switch 113 is provided in:.;.
the right plan view.
The pushbutton of the aforementioned slow; ;.:::
20 switch 104 is formed so as to be larger than the other;~
pushbuttons because the frequency of pushing of it is
higher. Further~ because the continuous slow pushing of
the pushbutton is tiring, the pushbutton is provided so
~hat it can be fixed. For example, there are used (1) a
slide lock type in which the pushbutton is locked when
it is pushed and slid laterally, (2) a double click type
in which the pushbutton is locked when it is clicked ..
twice, (3) a type in which the hold of the pushbutton is .
''`'''~"'
. ~ ~
, ~.
: :~
- 35 -
released when the reset pushbutton is pushed, and so on.
The aforementioned speech speed changeover
switch (speech speed setting switch) 113 is disposed
close to a range allowing the operation thereof with the
same finger so that this switch and the slow switch 104
can be operated alternately.
Besides the position in the aforementioned
embodiment, a ring switch, a slide switch, and so on,
may be used to make the operation easier.
- The aforementioned voice volume 108 is also
disposed in a range allowing the op~ration with the same
finger so as to be easy to adjust in order to always
make hearing in appropriate voice volume possible.
Further, a switch which has a feeling of soft
touch so that the microphone 321 does not pick up click
noise of the switch is preferably used as the afore-
mentioned switches high in the frequency of use, such as
the slow switch 104, the repeat switch 105, the reset
switch 106, the speech speed changeover switch 113, and
so on. For example, a switch using electrically
conductive rubber or the like is used.
Further, the external appearances of the
! aforementioned respective switches are preferably formed
into surface states which are different in the tactile
feeling in order to identify the switches in kind
without s~eing.
~ hen the aforementioned finger stop hollow 103
is opened, a structure is made so that some switches
.:. .. . . . , : ,. : . . .- . . . : . ., - .. .
- 36 -
such as a speech speed selection switch at repeat time,
and so on, axe seen.
The internal circuit structure of the speech
speed conversion apparatus in this embodiment is formed
so as to be identical to the aforementioned circuit
structure shown in Fig. 1.
As the PTL switch 4 in the previous embodi- ~-
ment, there are used the slow switch 104, the repeat
switch 105, the reset switch 106, and so on, as
10 described above. Furtherl as the selector switch 3 in ;;~
the previous embodimentl there is used the speech speed
changeover switch (speech speed setting switch) 113. `~
Further, the speech speed changeover switch (speech ;;
speed setting switch) 113 is connected to the external ; ;
interruption flag terminal 13 contained in the DSP 1.
The numeral value of the flag register 14 is changed by
the changeover of the speech speed changeover switch
113, so that the software 11 changes the expansion rate
of the speech speed conversion process in accordance
with this numerical value.
: i ;.:.
Fig. 11 is a block diagram showing the `~`
functional structure of ~he speech speed conversion ~ -
apparatus in this embodiment, in which the reference
numeral 21 designates speech input devices; 22, input ~ ~`
25 buffers; 23, a central processing unit (CPU); 24, a ring ~ ~-
buffer memory (which corresponds to the memory 2 in Fig. ;
1); 25, a function chooser; 26, output buffers; and 27,
speech output devices. ~
'' ,;. .
- 37 _
Making the constituent parts of this embodi-
ment correspond to those of the previous embodiment, the
speech input devices 22 are constituted by the micro-
phone 321, analog amplifier 10, low-pass filter 7 and
A/D converter 5 of Fig. 1.
The aforementioned input buffers 22 serve to
hold a speech converted into a digital signal by the
aforementioned speech input devices 21 and have a size
enough to hold data of the length of one frame which is
a unit for signal processing after that. These input
buffers 22 can be realized by the allocation of a part
of addresses of the ring buffer memory 24 ~which
corresponds to the memory 2 in Fig. 1).
The aforementioned central processing unit
(CPU) 23 which corresponds to the portion of software
executed on the DSP 1 shown in Fig. 1, has an encoder
23A, a silent-part elimination process 23B, a decoder
23C, a wave-form manipulation process ~speech speed
conversion process) 23D, and a controller 23E.
The aforementioned function chooser 25 which
corresponds to the portion constituted by the switches 3
and 4 and the external interruption flag terminal 13
shown in Fig. 1, is constituted by the slow switch 104,
the repeat switch 105, the reset switch 106, the speech
speed chanqeover switch 113, and so on, as described
above. .:
The aforementioned output buffers 26 which
serve to hold resulting data processed by the
- 3~ -
aforementioned wave-form manipulation process 23D are
two in practice and each of them has a size enough to
store data of the length of one frame expanded by wave-
form manipulation. In the previous embodiment, two
input buffers are provided so that a pipe-line process
is realized by using them alternately, whereas in this
embodiment a pipe-line process is realized by using two
output buffers alternately in the same manner as in the
previous embodiment.
That is, while the wave-form manipulation -
process of one frame is carried out so tha-t a result of
the process is held in one output buffer, a result o~
the wave-form manipulation process obtained in the
previous cycle is outputted from the other output buffer ;
via the speech output devices 27. These output buffers
26 can be realized by the allocation of a part of ad-
dresses of the ring buffer memory 24 (which corresponds
to the memory 2 in Fig. 1).
The inputting of data to the input buffers 22
and the outputting of data from the output buffers 26
are carried out at intervals of the sampling rate of the
A/D converter 5 and of the D/A converter 6 in the same
manner as in the previous embodiment. The process
executed by the DSP 1 is therefore constituted by a ;~
wave-form manipulation process for each frame and an
interruption process executed at sampling intervals. -
That is, the interruption process is executed
any number of times while the wave-form manipulation
,,. . . , . .. .. ,. .. , . ~ . . . . . . .
- 39 -
process is applied to data of the length of one frame,
so that the two processes are executed apparently and
simultaneously~
As the aforementioned ring buffer memory 24,
5 there is used a well-known type memory in which
writiny/reading is performed for each frame. The
details thereof will be described below.
(Writing Operation)
In Fig. 11, speech data inputted through the
speech input devices 21 are held in the input buffers
22. The input buffers 22 have a capacity enough to hold
a number of data corresponding to one frame so that the
code length of 16 bits per one data is allocated
thereto, and the input buffers 22 are realized by the
allocation of a part of addreæses on the memory 2 shown
in Fig. 1.
The controller 23E shown in Fig. 11 monitors
the state of these input buffers 22 and transfers speech
data of the length of one frame to the encoder 23A
whenever the input buffers 22 are filled with the data
of the length of one frame.
In the encoder 23A, the input speech data of
the length of one frame is subjected to an information
compression process, so that the data as a result of the
compression is held in the ring buffer memory 24.
Several methods are considered as this compression
process. One example thereof is a difference data
holding method shown in Figs. 12A and 12B. Figs. 12A
. - . . ... -.,: ,. , . ., ~. .. .. . .
- 40 -
and 12B are typical graphs for explainin~ the compres-
sion process in the encoder 23A in this embodiment.
In this compression process, "difference from
the previous data" is calculated successively from the
leadiny data of each frame. In Fig. 12A, the~e differ-
ence data are expressed as ~1, Q2,... The output data
of the compression process are data obtained by arrang-
ing the aforementioned difference data ~ 2~ into
the code length of 8 bits per one data after dividin~
10 the leading data of the frame into upper 8 bits and ``~
lower 8 bits. One data of the input data has a digital
code length of 16 bits. In the case of an input signal
such as a ~oice signal which changes sufficiently slowly `
compaxed with the sampling interval, the difference from "~
15 the previous sampling value is however not so large that ; `~
the difference can be e~cpressed sufficiently in the code -
length of 8 bits which is a half as shown in Fig. 12B.
The capacity of data after the compression process is ~``
therefore about a half the capacity of data before the
compression pxocess but there is no missing from the
contents thereof as long as the difference in the middle
of the process does not become too large to be expressed
in the code length of 8 bits.
In the storage into the ring buffer memory 24,
data thus compressed into a half capacity for each frame
~re arranged on the ring buffer memory 24 so that the
time sequence thereof is maintained.
In addition to this, a frame header is added
:
- 41 -
to the leading of the compressed data o each frame in
order to indicate a break between frames. In the
compression process portion, the calculation of the sum
of the absolute values of all data in the frame as well
as the aforementioned compression process in Fig. 12 is
carried out and, at the same time, the work of recording
a result thereof as the power value of this frame in the
aforementioned frame header portion is carried out.
The determination of a frame to be subjected
to the waveform expansion/reduction process is performed
on the basis of comparison between the power of the
frame and the threshold Th. Further, the silent-part :
elimination process is carried out on the basis of
comparison between the power of the frame and the
threshold To.
It is preferable that these thresholds are not
used as fixed values but changed in accordance with the
loudness of the input voice. For example, between the
case of use in a quiet room and the case o~ use in a
situation of large background noise, speech speed con-
version, of course, cannot be performed well unless
these thresholds are adjusted well.
- In a specific realization method, the maximum/
minimum values of frame power in the past period of
several seconds are stored so that the aforementioned
thresholds are determined on the basis of these values.
For example, in the case where these thresholds are to
be changed at intervals of five seconds in the condition
" ' : ' ' : ' ' , ' ' .: , '
- 42 -
in which the time length of one frame is 50 milliseconds
(msec), the process of changing the threshold Th can be
carried out once whenever 100 frames are processed.
As described above, the power of each frame is
always calculated with respect to all inputs whenever
information compression is performed for each frame by
the encoder in Fig. 11, so that information thereof is ;~
recorded in the frame header and held in the ring buffer ;`~
24. `-~
In this calculation of frame powex, the `~
maximum frame power Pmax and the minimum frame power`~
Pmin are compared with each other so that they are
updated if necessary. If the maximum frame power Pmax
and the minimum frame power Pmin are provided so as to
15 be re~et at intervals of five seconds (100 frames), the`~
maximum frame power and the minimum frame power in the
past period of five seconds can always remain.
In the calculation of the thresholds, for
example, Th and To are set to 10 % and 5 % the differ-
ence between the maximum frame power Pmax and theminimum frame power Pmin, respectively. These are given
by the following expressions (1) and (2).
Th = ¦Pmax - Pmin¦*0.10 + Pmin (1) ~ -
To = ¦Pmax - Pmin¦*0.05 + Pmin (2) ;
' '':: ' "
Although the method of holding raw data in the
ring buffer memory 24 in this embodiment has been
''` :'
- ~3 -
described above, the details of silent-part elimination
will be described below.
As explained in the previous embodiment with
reference to Fig. 3, the function of silent-part elimi-
nation serves to eliminate a silent part (a duration inwhich power is lower ~han the voice-part/silent-part
threshold To) continued for a time not smaller than one
secondO
The silent-part elimination process is carried
out by the silent-palt elimination process 23B shown in
Fig. 11. This silent-part elimination process is a
process independent of a later-described process
executed for each frame ~hereinafter referred to as a
main process) so that the process is carried out after
the main process for one frame is terminated. In Fig.
14, the process is carried out between the judgment
"delay=0?~ (S143) and the ~udgment "Power ON?" (S144)
(though not shown).
In the silent-part elimination process 23B,
data accumulated in the input buffers 22 are added up at
intervals of a predetermined unit (for example, 1/4
frame) to calculate power, so that the silent-part
elimination operation is started when the power "cxosses
the voice-part/silent-part threshold upwards". This is
because the point of time of termination of the silent
part is the point of time of the change of power from a
small value to a large value and, at any point of time
except this point of time, a judgment cannot be made as
- 4~
to whether the silent part continued up to that is ~-
longer than one second or not~
When the silent-part elimination process is
started, first the frame heacler of the ring buffer
memory 24 is retrieved retroactively to the past.
Compressed data on the ring buffer memory 24 are com-
pressed for each frame and, clS described above, the
power value of the frame is recorded in the frame
header. If a frame having power lower than To is
continued for a time not smaller than one second,
silent-part elimination is enabled and the input pointer
to the ring buffer memory 24 is returned to the point of
time in which the silent part has been continued for one
second. The input of the next compressed data is
~:.. .
recorded so as to be o~erwritten from the returned point
of time. ~ccordingly, the silent part continued for a
.. .. .
time not smaller than one second just before the current
point of time is always eliminated.
;:":. ' "'
(Reading Operation)
The later-described main process in the
apparatus of this embod.iment is carried out for each
frame. The wave-form manipulation process 23D shown in
! Fig. ll therefore holds currently processed frame data, ~ -
so that reading from the ring buffer memory 24 is
performed collectively for each frame. That is, because
addressing to the ring buffer memory 24 can be made
easily by a process of increasing the address one by one
simply in the case where data are collectively picked
;: :
'~' "
- ~5 -
out, this case is better in efficiency than the case
where data are one by one picked out.
Because the data stored in the ring buffer
memory 24 are compressed data as described above, it is
necessary that this compression is decoded into the
original data before the wave-form manipulation process.
The decoder 23C shown in Fig. 11 is provided for this
purpose. First, leading two 8-bit data are arranged in
the uppertlower of 16 bits with one-frame compressed
data as an input to ~enerate a leading data. Then, the
value of the third data of the compressed data is added
to the leading data to restore the second data. Then,
the value of the next data of the compressed data is
added to the second data to restore the third data.
Thereafter, the work of adding the compressed data to
the previously restored data successively is repeated
thus to restore all data of the frame.
The basic operation of the speech speed
conversion apparatus in this embodiment will be
described below in brief.
As shown in Fig. 11, the speech converted into
a digital signal by the speech input devices 21 is first
inputted to the input buffers 22. The speech signal
read from the input buffers 22 is fed to the encoder 23A
contained in the CPU 23 of DSP1 (Fig. 1), subjected to
the data compression process and stored in the ring
buffer memory 24. The aforementioned speech signal is
also fed to the silent-part elimination process 23B so
... , ... . . . , . ~; ~ ,, :
- 46 -
that the silent-part elimination process is applied to
the data stored in the ring buffer memory 24 if
necessary.
The data of the speech signal stored in the
ring buffer memory 24 are frame-by-frame fed to the
decoder 23C, so that the compressed speech data are
decoded by the decoder 23C and inputted to the wave-form
manipulation process (speech ~peed conversion process) .' ,
23D. In the wave-form manipulation process (speech. .:
10 speed conversion process) 23D, there is carried out :~
speech speed conversion or the like on the basis of the -
condition set by the function chooser 25. ~he digital -`
speech data subjected to the speech speed conversion` ;
process or the like are held in the output buffers 26.
15 The data of the output buffers 26 are read out so that:~
the speech subjected to the speech speed conversion::
process or the like is outputted from the speech output
de~ices 27.
That is, the data of the output buffers 26 are
read out so that the data are converted from a digital
value into an analog value at intervals of a set time by ; .
the D/A converter 6 as shown in Fig. 1. The analog .
. , .
signal thus obtained by this conversion is inputted to ; :
the analog amplifier 9 via the low-pass filter 8 and
outputted clS a voice from the binaural headphone 325 in
listener's favorite amplitude of speech signal.
Referring to Figs. 11, 13 and 14, the process
executed for each frame (hereinafter referred to as a
,~
' . . , i !' , , ' . , . . " ' ;.:. . ! . .
- ~7 -
main process) in this embodiment will be described
below.
Figs. 13 and 14 are flow charts showing the
procedure of the main process in this embodiment.
As shown in Fig. 13, in the main process in
this embodiment, the "fade-in" step is carried out
~S131) with Powering ON. That is, just after the
powering-on of the electric source, data stored in the
output buffers 26 are indefinite. Just after the
powering-on of the electric source, data having no
relation to the speech may be therefore outputted. In
the case where the data are outputted intact from the
speech output devices 27, the data may form noise of a
very large level. To prevent this, in this embodiment,
the values of data in the output buffers are adjusted by
the execution of the fade-in step so that the output of
the speech output devices is increased gradualIy for a
predetermined time after the powering-on of the electric
source irrespective of the data in the output buffers.
Specifically, whenever one data is transferred from the
output buffers to the D/A converter, the ~alue of this
data is multiplied by a coefficient, so that this
function is realized by changing the value of the co-
efficient with the passage of time. This operation is
executed by the controller 23E shown in Fig. 11.
Thereafter, the "through mode" process is
started. In the through mode process, first, the
"reading pointer coincidence" step is carried out
- 48 -
(S132). This reading pointer coincidence process is a
process in which when data from the speech input devices
21 is inputted, the same data is inputted to the output
buffers 26 just after the inputting of the data to the
5 input buffers 22. This operation is realized by making ; .
, .. . .
the value of the input pointer pointing an input address .. :
on memory coincident with the value of the output -
pointer pointing an output data address on memory just .
after the inputting of data to the input buffers 22. In .
Fig. 11, this operation is carried out by the controller
23E. :~-
After the reading pointer coincidence, in the
through mode, the pushed states (ON states) of the slow
switch 104 and the repeat switch 105 are checked (S133
and S144). In the case where both switches are in non-
pushed states (OFF states), the situation of the routine .
goes back to the previous reading pointer coincidence
step (S132) so that the through mode is continued.
Accordingly, in the interruption process which occurs
while the through mode is continued, input data isalways outputted intact, so that the same speech as the
input speech is outputted from the speech output devices
27.
In Fig. 11, the aforementioned respective .
switches such as the slow switch 104, the repeat switch
105 and the reset switch 106 are contained in the ;
function chooser 25 and the states thereof are checked
by the controller 23E.
- 49 -
When the repeat switch 105 is pushed (turned
ON) in the aforementioned through mode, a repeat flag
(not shown) prepared separately is set from 0 to 1 and
the ~reading pointer return~ routine is carried out
(S135). Fig. 16 shows a flow chart of the internal
procedure of this reading pointer return routine. The
explanation of Fig. 16 will be described later.
When the slow switch 104 is pushed in the
aforementioned through mode, the situ tion of the
routine skips to the routine for "setting parameter in
expansion process" (S136) as shown in Fig. 13. Fig. 17
shows a flow chart of the internal procedure of this
routine. The explanation of Fig. 17 will be described
later.
After the setting of parameter in the
expansion process is performed, the one-frame waveform
expansion/reduction process is carried out (S137).
Figs. 18 and 19 show flow charts of the internal
procedure of this one-frame waveform expansion process.
The explanation of Figs. 18 and 19 will be described
later.
After the aforementioned one-frame process is
completed, the situation of the routine goes to the step
for checking the states of the respective switches as to
whether each switch is pushed or not. Hereupon, because
the one-frame process is terminated within the time
length of one frame, the process is completed in the
oxder of tens of milliseconds (msec). On the other ~
' " ': ': '
- 50 -
hand, switching devices which are such that the pushed
states are maintained for a time not shorter than the
duration of pushing, no matt,er how short, in the case
where the respective switches (pushbuttons) are pushed
by a user, are used in this apparatus. Accordingly, the
situation of the routine can be shifted to a desired
operation with such a time lag that a feeling of slow
response is not given to the user, as long as the pushed
states of the ~witches are checked whenever the one-
10 frame process is carried out. ~ ~
First, whether the reset switch 106 is pushed `
down or not is checked (S138). If the reset switch 106
is pushed down (in the case of Yes in S138), the current
mode forcedly goes to the through mode at this point of
time.
If the reset switch 106 is not pushed down (inthe case of No in S138), whether the slow switch 104 is
pushed down or not is checked (S139) as shown in Fig.
14. If the slow switch 104 is pushed down (in the case
of ON: in the case of Yes in S139), the situation of
the routine goes back to the routine in which parameter
is set in the sxpansion process so that the wave-form
expansion process is applied to the next frame continu-
ously. In the case where the slow switch 104 is pushed
down continuously, the situation of the routine circu-
lates in this loop continuously. Further, even in the
case where the slow switch 104 is pushed down continu-
ously durin~ the repeat reproduction and ca~ching-up
- 51 -
reproduction, the situation of the routine circulates in
this loop continuously.
In the case where the slow switch 104 is
opened (in the case of OFF: in the case of No in S139),
the situation of the routine goes to the next judgment
as to the repeat pushed-down state (S140~. The case
where the pushing-down of thle repeat switch 105 is
detected at this point of time is either the case where
~the repeat switch is pushed at the time of repeat
reproduction" or the case where "the repeat switch is
pushed at the time of catching-up reproduction~. In
either case, the situation of the routine branches into
the reading pointer return routine so that the repeat
reproduction is started from the silent part near a
15 position returned back to the past by about five seconds -~
from the current position of the output pointex of the
ring buffer memory 24.
In the case where the slow switch 104 is
opened and the repeat switch 105 is not pushed down, the ~:
20 situation of the routine goes to the following repeat ~. :
end judgment tS141). The repeat operation is continued
until the output pointer goes bacX to the output pointer ::: ~
position where the through mode was changed to the .`;
repeat operation by the pushing-down of the repeat
switch 105. That is, in the case where this judgment
shows that the repeat mode is used currently and that ~.
the position of the output pointer does not yet go back
to the output pointer position where the repeat was .;;
- 52 -
started, a processing loop is formed so that the
situation of the routine goe~; back to the aforementioned
one-frame waveform expansion/reduction process. The
subsequent process is a process for catching-up
reproduction.
After the repeat reproduction is terminated or
after the slow reproduction is terminated, the situation
of the routine goes to the catching-up reproduction.
The catching-up reproduction means an operation in which
a time lag from the real time as caus~d by the repeat or
slow reproduction is made up for by fast reproduction
realized by the repetition of the one-frame waveform
reduction process. In the process in this portion, the
setting of parameter is performed for the waveform `
reduction process for the catching-up reproduction
(S14~).
The quantity of the lag from the real time
increases when the repeat button is pushed down or when
the waveform expansion process is carried out. On the
20 contrary, it decreases when the waveform reduction "
procass is carried out.
The process for increasing~decreasing this
quantity of the lag is however not shown in Figs. 18 and
19 which show the procedure ~flow chart) of the one-
frame waveform expansion/reduction process which will be~escribed later.
A judgment is made as to whether the quantity
of the lag from the real time is present or absent
,.. . .. .- .. , . . , . . ~ ., , ~ , . . . . , . , . .. , ., , , , . . , ., ~
- 53 -
(S143). In the case where the quantity of the time lag
is present yet, a processing loop is formed so that the
catching-up reproduction is continued. That is, the
operation in which the catch:ing-up reproduction is
continued until the quantity of the time lag becomes
zero, is realized by this judgment.
On the other hand, in the main process '
described above, the time lag from the real time as
caused by the speech speed conversion or repeat
operation is managed as "lag quantity" by using a
counter. ' ;
Although the time lag from the real time can -
be managecl also as difference between the position on
the ring buffer 24 where the current sampled data is
inputted and the position on the ring buffer 24 where
the position of data outputted is inputted, that is, as
difference between addresses pointed by two pointers,~; '`
the management method using the lag quantity counter as
described above is employed in the present invention.
20 This is because the quantity of the lag may be unable to ~ ~;
be expressed correctly in the address difference between '
the input and output pointers on the ring buffer 24.~;;; ;
'~ For example, assuming that the memory a'ddress
space allocated to the ring buffer 24 is from address 0
to address 1000, then the ring buffer 24 is realized by
the handling of the memory address space in a manner of
"next to acldress 1000, jump to address 0" in the ~;
program. Therefore, in the case where the input and ''`~.
,' ''~:
~.'",-'
- 54 -
ou~put pointers lie across this break between addresses,
the quantity of data therebetween cannot be expressed
easily by taking the difference between address values
simply. In order to known the quantity of data between
these pointers by address ca:Lculation, address value
calculation including complex classification that takes
into account the histories of the two pointers up to :
their current positions is required.
In the speech speed conversion apparatus
according to the present invention, whenever reading~
writing of data is performed with respect to the rin~
buffer 2~, the value of ~he lag quantity counter is
chan~ed to thereby manage the quantity of the time lag
to prevent the increase of the quantity of processing
based on the complex address calculation.
The aforementioned main process is provided in ~
the form of an infinite loop in which the aforementioned :
process is repeated until the electric source switch is
turned o~f (S144).
In the case where the electric source switch
is turned off, the process is not suddenly stopped but
continued for a predetermined time before it is stopped
(mute) (S145). During this time, here is carried out
such a process that the loudness of the output voice is
reduced gradually.
Specifically, in an interruption process in
the same manner as in the fade-in operation which is the
first step, whenever one data from the output buffers 26
- 55 -
is transferred to the D/A converter 6 shown in Fig. 1,
the value of this data is multiplied by a coefficient so
that this function is realized by changing the value of
this coefficient with the passage of time. This
operation is carried out by the controller 23E shown in
Fig. 11.
Fig. 15 is a state transition view typically
showing transition between respective modes in this
embodiment as described above. The way of mode switch-
ing on the basis of the switching operation will be
... ... .
understood well from Fig. 15. Further, the standby mode
in Fig. 15 will be described later in detail.
The details of processing operations in the ~
respective routines in this embodiment described above ~;
will be described below in detail.
Fig. 16 is a flow chart showing the procedure ~;
of the reading pointer return routine.
The reading pointer return routine in this
embodiment is a specific method for changing the value
of the output pointer pointiny the position of data to
be read from the ring buffer 24, which method is
necessary for realizing a repeat function.
As shown in Fig. 16, first, the position of
the output pointer at the current point of time is set
to Pout (S161). Then, the quantity of the lag from the
real time at the current point of time is set to D .
(S162).
A judgment is made as to whether the quantity
., . -~ ~ . ..
.:: , ', '~.
.~ '`. '
of the lag is already large at the current point of time
so that the quantity of the lag will exceed the size of
the ring buffer memory 24 if the quantity of the lag is
further increased by 5 seconds (B5) (S163). In the case
where a decision is made as ,a result of the judyment so
that the quantity of the lag exceeds the size of the
ring buffer memory 24 (the case of Yes in S163), this
routine is terminated without any change of Pout and D
(S169 and S170).
In the case where the quantity of the lag can
be increased by five seconds ~B5) (the case of No in
S163), the pointer i~ returned back by five seconds
(-B5) and the quantity of the lag is increased by five
seconds (+B5) (S164).
Then, a process of searching back for the
silent part is started so that the start of the repeat
is made a pause of the speech. First, data is accessed
backward from the position pointed by Pout on the ring
buffer memory 24 to thereby calculate one-frame power
(S165).
At this time, if the output pointer is
returned back (-F) by one frame (F), the quantity of the
lag is also further increased by one frame (+F). Here,
a judgment is made as to whether or not the total quan-
tity of the lag exceeds the size of the ring buffer
memory if the quantity of the lag is further increased
by one frame (S166), and in the case where a decision is
made as a result of the judgment so that the total
... ~
- 57 -
quantity of the lag exceeds the size of the ring buffer
memory (the case of Yes in S166), this search for the
silent part is stopped and Pout and D at this time are
set as the output pointer vale and the lag quantity
respectively (S169 and S170) whereafter this routine is
terminated.
In the case where the total quantity of the
lag does not exceed the size of the ring buffer memory
(the case of No in S166) though the output pointer is .
returned back by one frame, Pout is returned back by the
length of one frame and the quantity D of the lag is
increased by one frame (S167) whereafter the calculated
one-frame power W and the voice-part/silent-part thresh-
old are compared with each other (S168). In the case
where the one frame power W is smaller than this thresh-
old, a decision is made that a pause of the speech is
present near this frame (the case of No in S168) and
Pout and D at this time are set as the output pointer
value and the lag quantity respectively (S169 and S170)
20 whereafter this routine is terminated. ~`~
In the case where the one-frame power W is ;~
larger than this threshold (the case of Yes in S168), ~-
the pointer is further returned back by one frame and
the search for the silent part is continued to detect
the silent part in the same manner as described above
but the search is continl7ed until the quantity of the ~ ;
lag exceeds the size of the riny buffer memory. Thus, -
the output pointer return process at the time of the~ ~
'',." . '~
~'`' ~'''
- 58 -
pushing of the repeat switch 105 is completed.
Figs. 17 and 18 are flow charts showing the
procedure of the one-frame waveform expansion/reduction
process in this embodiment.
In the one-frame waveform expansion/reduction
process in this embodiment, as shown in Figs. 17 and 18,
first of all, power of current one-frame data is calcu-
lated (S171). Then, this power value P is compared with
the threshold Th (S172). A frame having higher power
than the threshold Th is subjected to the following
process. Data in a frame having lower power than the
threshold Th may be outputted intact to be transferred
to the ring buffer 24 (S173) or may be subjected to a
consonant emphasis process and then transferred to the
output buffers 26. Whether consonant emphasis is to be
performed or not to be performed is determined by the
state of the mode switch which is one of hidden
switches.
As a specific method for realizing the
consonant emphasis process, there is, for example,
considered a method in which a frame having lower power
than the threshold Th just prior to a frame having
higher power than the threshold Th is regarded as a
consonant and the values of data in the frame are
increased.
In the case of a frame having higher power
than the threshold Th in the aforementioned power
judgment (the case of Yes in S172), first the number of
.. .~ . . .- - . : .,:., - - ,
- 59 -
data in one frame is stored in a variable Z indicating
the quantity of not-yet-processed data (S174) and then
the pitch extraction process is carried out from the
leading of the frame (S175). Several methods are
considered as the pitch extraction process. For
example, the pitch length at the leading of the frame is
extracted on the basis of a well-known algorithm using
autocorrelation.
Then, the quantity of data corresponding to
twice the thus extracted pitch length is compared with
the quantity of not-yet-processed data (S176), and in
the case where the quantity Z of not-yet-processed data
is smaller the quantity of data twice as much as the `~
extracted pitcht this process is stopped.
In the case where the quantity Z of not-yet-
processed data is equal to or more than the quantity of
data twice as much as the extracted pitch (the case of
Yes in S176), a pre-transfer process is carried out
(S178). The pre-transfer process means a process in
which a part of input data is transferred intact to the
output buffer 26 before a reproduced wave pattern ~`
insertion process which will be described later. The
pre-transfer process corresponds to the portion of (b)
in Fig. 2. The number of data to be transferred by the ;
pre-transfer process is set with the pitch as A unit but
the number thereof varies in accordance with the wave-
~ .,.~ .
form expansion/reduction rate. The number Npf is set
(S177) by a parameter setting routine which will be ~ -
" '
, ~ .
.i~-`~, `" '
- 60 -
described later with reference to Fig. 19. After the
pre-transfer process is carried out (S178), the quantity
Z of not-yet-processed data is reduced by the number of
transferred data (S179).
Then, the position of application of a Q
window function for generating a reproduced wave pattern
is determined (S180) in accordance with another para-
meter Ptri set in the parameter setting routine shown in
Fig. 19. What differs between expansion and reduction
is only the position on current wave to which the window
function is applied in the case where a reproduced wave
pattern is generated by using the ~ window function.
That is, in the case of waveform expansion, as
shown in Fig. 2, the ~ window function is applied so
that waveform of the length of two pitches is generated
from waveform of the length of one pitch (S181~.
Contrariwise in the case of waveform reduction, as shown
in Figs. 20 to 22, the ~ window function is applied so
that waveform of the length of two pitches is generated
from waveform of the length of three or four pitches.
The quantity of the lag from the real time is changed by
the insertion of the reproduced wav0 pattern (though not
shown).
After the reproduced wave pattern insertion
process, the quantity Z of not-yet-processed data is
reduced by the number of the thus processed data (S182).
llhen, the pitch extraction process is carried
out again (S183). This is a process which is adapted to
:, . . . .
,...... ~ . . ~ .. . . , ~ . .:
. ' ~ ' , ` , ' ; ~
-: . ; . ~ , , j ,
: . . : . -
- 61 -
the fact that the human voice pitch always varies and in
which the error between the aictual pitch length and the
pitch length for processing is reduced by extracting the
pitch again to thereby consequently prevent the increase
of distortion in waveform after expansion/reduction.
Then, as shown in Fig. 18, the quantity of
data twice as much as the ne~wly extracted pitch is
compared with the number of not-yet-processed data
(S184). If the quantity of data of the length of two
pitches does not remain (the case of No in S184), this
process is stopped immediately.
If the quantity of data larger than the length
of two pitches remains (the case of Yes in S184), a
post-transfer process is carried out. The post-transfer
process means a process similar to the pre-transfer
process and corresponds to the portion of (e) in Fig. 2 ~
in the previous embodiment. The number of data to be ;
transferred by the post-transfer process is set with the
pitch as a unit but the number thereof varies in accord~
ance with the waveform expansion/reduction rate. The
number Npf is set (S185) by the parameter setting -`
routine which will be described later with reference to
Fig. l9. After the pre-transfer process (S186), the ,
quantity Z of not-yet-processed data is reduced by the
25 number of transferred data (S187). -
The aforementioned procedure is continuously
repeated until this procedure is stopped on the basis of
comparison between the quantity of data of the length of ~-~
- 62 -
two pitches and the quanti~y of not-yet-processed data
which comparison is performed twice in the middle of
this procedure.
Fig. 19 is a flow chart showing the procedure
of the parameter setting routine for setting parameter
for the expansion process in this embodiment.
In practice, the parameter setting routine
shown in Fig. 19 is used twice in the main process shown
in Figs. 13 and 14. Once thereof is used just before
the aforementioned one-frame waveform expansion/reduc-
tion routine and the other once is used in a "process
for setting parameter for the reduction process" after
the repeat end judgment.
Hereupon, the waveform reduction process is a
process for realizing the "catching-up process (fast
hearing process)" which is continued after slow hearing
or ater repeating. When the generation of a reproduced
wave pattern with use of the ~ window function as car-
ried out in the waveform expansion process is carried
out while the position subjected to the window function
is shifted in a direction reverse to the case of
expansion, waveform reduction is obtained.
In Fig. 1~, first a discrimination is made
between expansion and reduction (S191). This discrimi-
2~ nates one of the aforementioned twice from the other.
In the case of parameter setting for theexpansion process, after this discrimination, the
position of the speech speed selection switch is checked
- 63 -
(S192), the expansion rate e is set in accordance with ~
the position of the switch (S193), the positions of ~-
parameters Npf and Npr used in the waveform expansion
process are set in accordancle with the expansion rate e, ;
and parameter Ptri indicating the position of the start
of weighted summation with r~espect to the ~ window as
carried out in the waveform expansion process is set,
whereafter this routine is terminated.
On the other hand, in the case of parameter
10 sstting for the reduction process, the right flow in ~;
Fig. 19 is carried out. First, the position of the
catching-up mode switch twhich is one of hidden
switches) is checked (S196) and which of "jump", ~'fast
hearing" and "one-fold" the catching mode (Mcat) is set
to is chècked (S197 and S198).
When set to "jump", the catching mode (Mcat) ;~
practically serves not to "catch up" but to jump
actually just at the moment that the hold of the slow ,
switch (slow pushbutton) is released (S199). Specifi-
20 cally, a branching process for forcedly returning back ;~
to the through mode is carried out in this portion.
When the catching-up mode ~Mcat) switch is set
to "one-fold", the reduction rate s is set to one-fold -
(S200) and the situation of the routine goes to step
202.
When the catching-up mode (Mcat) switch is not
set to "one-fold", the reduction rate s is set through
the center flow in Fig. 19 at the time of the ca~ching-
'", :, : ., -: . ' . ~ ' , . ,., , " , , . ' : . . ' .. ' ' . ' . , .
- 64 _
up mode (S201), the values of parameters Npf and Npr
used in the waveform reducti.on process are set in
accordance with the reduction rate s (S202) and,
further, parameter Ptri indicating the position from
which weighted summation wi~h respect to the ~ window as
carried out in the waveform reduction process is started
is set (S203), whereafter this routine is terminated.
Figs. 23 and 24 are flow charts showing the
procedure of the total operation of a speech speed
conversion apparatus provided with a continuous speech
speed conversion means according to the present ..
invention.
As shown in this embodiment of Figs. 23 and
24, continuous speech speed conversion in the speech
speed conversion apparatus provided with the continuous
speech speed conversion means is substantially an
operation in which the pushing of the slow switch (slow
pushbutton) 10~ is continued so that slow reproduction
is continued. The time lag is however accumula~ed
rapidly when waveform expansion at a constant waveform
expansion rate is continued, so that the ~uantity of the
lag from the real time finally exceeds the capacity of
the ring buffer 24 to make it impossible to continue
slow hearing any more.
The continuous speech speed conversion means
is therefo:re provided to mix a waveform expansion period
and a waveform reduction period reverse the.reto at the
time of slow reproduction so that the lag from the real
- 65 -
time is not increased rapidly.
Although several methods are considered as
means for changeover to the continuous speech speed
conversion mode, it is rathe:r easy to understand that a
clear distinction is made between the case where the
pushing of the slow switch is continued simply for a
considerably long time and the case where entry into the
continuous speech speed conversion mode is intended.
Accordingly, this changeover is realized , for example,
by using switching parts by which the slow switch is
locked when double-clicked (pushed twice at a shoxt time
interval) or when slid laterally while pushed.
The respective steps in the flow charts shown
in Figs. 23 and 24 in this embodiment are quite the same
as in the procedure of the main process described above
with reference to Figs. 13 and 14.
In the continuous speech speed conversion
means in this embodiment, whether the continuous speech `
speed conversion process is intended or not is checked
in step S231 in Figs. 23 and 24 (S231). If the continu-
ous speech speed conversion process is intended (the
case of Yes in S231), the one-frame waveform expansion/
reduction process is carried out (S232). Then, a
judgment is made as to whether the reset switch 106 is
pushed (turned on) or not (S233). In the case where the
reset switch 106 is not pushed (turned off), counting up
by one frame is performed (S234) and a judgment is made
as to whether the expansion period is intended or not
. . .
- 66 -
(S235). If the expansion period is intended (the case
of Yes in S235), the situation of the routine goes back
to the step S232. If the expansion period is not
intended (the case of No in S235), parameker is sek for
the reduction process (S236). Then, whether the lag
quantity is zero or not is checked (S237). In the case
where the lag quantity is zero (the case of Yes in
S237), the situation of the routine goes back to the
step S232. In the case where the lag quantity is not
zero (the case of No in S237), parameter is set for the
expansion process (S238) and the frame counter is reset
(S239) whereafter the situation of the routine goes back
to the step S232 so that the continuous speech speed
conversion operation is repeated. In the case where the
continuous speech speed conversion process is not
intended in the aforementioned step S231 (the case of No
in S~31), the mode is shifted to the aforementioned main
process routine (through mode).
That is, the continuous speech speed conver-
sion means in this embodiment is a method in which slow
reproduction and catching-up reproduction are repeated
alternately at intervals of a preliminarily set time.
According to this method, catching up to the real kime
at inter~als of a predetermined time is always made
possible. The management of the changeover between
waveform expansion and waveform reduction is performed
on the basis of the count of the number of fxames. For
example, when the expansion process for a number of
- 67 -
frames corresponding to about five seconds is completed,
the r~duction process is the:n carried out repeatedly, -
and when the lag quantity reaches zero, the frame count
is returned to zero and the ~expansion process is
5 repeated again. ~-
Further, escape out of the continuous speech
speed conversion mode is achieved by the pushing-down of
the reset switch 106 to return the mode to the through
mode.
Referring next to Figs. 25 and 26, there are ~:
flow charts showing the procedure of the total operation .~ ;
of a speech speed conversion apparatus provided with a
continuous speech speed conversion means different from
that in the embodiment shown in Figs. ~3 and 24.
The continuous speech speed conversion in the
speech speed conversion apparatus provided with the ;~
continuous speech speed conversion means in this embodi-
ment is an operation for applying waveform expansion to
a frame of high power and applying waveform reduction to .:
a frame of low power.
In the continuous speech speed conversion ..
means in this embodiment, whether the continuous speech
speed conversion process is intended or not is checked
in step S251 in Figs. 25 and 26. If the continuous :
speech speed conversion process is intended (the case of
Yes in S251), a judgment is made as to whether the reset ; .
switch 106 is pushed (turnPd on) or not (S252). In the
case where the reset switch 106 is not pushed (turned
- 68 -
off), one-frame power is calculated (S253). Then,
whether the calculated one-frame power is higher than
the threshold Th or not is checked (S254). In the case
where the calculated one-fra:me power is lower than the
threshold Th (the case of No in S251), parameter is set
for the reduction process (S256) and the situation of
the routine goes to step S257. In the case where the
calculated one-frame power is highex than the threshold
Th (the case of Yes in S2~4), parameter is set for the
expansion process (S255) and the one-frame waveform
expansion/reduction process is carried out (S257) where-
after the situation of the routine goes back to the step
: S252 so that the continuous speech speed conversion
operation is repeated. In the case where the continuous
speech speed conversion process is not intended in the
aforementioned step S251 (the case of No in S251), the
mode goes to the aforementioned main process routine
(through mode~
That is, when entry into the continuous speech : `
speed conversion mode is made, one-frame power is
calculated so that either expansion or reduction is
applied to each frame on the basis of comparison between
the one-frame power and the threshold Th. Escape out of
the continuous speech speed conversion mode is achieved ..
by the pushing-down of the reset switch 106.
In this embodiment, the speech is made slow or
fast in accordance with the power thereof.
In ordinary conversation, there is generally a
','"",;. ',
: '',~",,''~;'.'
:
- 69 -
tendency that an important portion which must be told ~o
a listener is louder-voiced but a portion not so import-
ant is smaller-voiced. Accordingly, the speech speed
control in this embodiment is characterized in that an
output voice nearer to the natural voice is obtained.
The probability of appearance of the high-
power portion and the probability of appearance of the
low-power portion are however not always equal to each
other, so that catching up to the real time at intervals
lG cf a predetermined time as in the case of the previous
embodiment of Figs. ~3 and 24 is not always ensured.
Further, as a method of instruction from the
user to attain entry into the continuous speech speed
conversion mode, there are considered a method in which
the slow switch (slow pushbutton) 104 is pushed and then
slid laterally to thereby be locked, a method in which
the slow switch (slow pushbutton) 104 is double-clicked
(pushed down twice in succession at a short time
interval), and so on. If these methods are used, the
respective intentions of "executing slow reproduction"
and of "continuing" the operation by the pushing of the
slow switch (slow pushbutton) 104 can be expressed in
difference in the way of pushing of the same pushbutton
so that there can be provided an operating system which
is more intuitive and easier to understand compared with
the case where a continuous speech speed conversion
pushbutton is provided separately.
~rhe embodiments up to now have been described
- . :
- 70 -
above upon the assumption that the waveform ~expansion
rate" in the case of "slow" reproduction based on the
waveform expansion process is determined by the setting
of the "speech speed settiny switch" provided on the
apparatus and that a "default value~ determined (in the
program) in advance is used as the waveform ~'reduction
rate" in the case of "fast'l reproduction based on the
waveform reduction process.
The function of ~making freely coming and
soing on the time axis of the speech possible" provided
by this apparatus, however, can be used by the user more
intuitively when an "accelerator type switch" shown in
Fig. 27 is provided so that the waveform expansion/ -~
reduction rate is changed by this switch.
When the accelerator type switch i~ set to the
center, the through mode in the aforementioned embodi-
ments is executed. When the slide switch is pulled to
the front, the waveform expansion process is employed so
1 .,
that "slow reproduction" is executed with a lag from the
20 real time. When the slide switch is then pushed to the ~-
: ~ ;
back, the waveform reduction process is employed
reversely so that fat reproduction is execute~ (until -;~
the lag from the real time reaches zero).
During this, the controller changes the ~
25 waveform expansionJreduction rate in accordance with the ~ '
distance from the center of the slide switch. As is
obvious from the explanation of the aforementioned
embodiment of Figs. 20 through 22, the expansion/reduc- ~;
:.'' '',~-;':
'~ '~''' .''
- 71 -
tion rate however can be set to no value but several
values which can be expressed in integer rates. In
practice, therefore, the expansion/reduction rate may be
preferably set so that several stages of values can be
selected in accordance with the distance from the center
of the slide switch.
Further, when a lever is provided so that
force acts to return the lever to the center automati-
cally when the user releases the finger's hold of this
accelerator type switch, it becomes easy for the user to
keep the slide switch in another intermediate position
than the center, sc that an operating method easier to
handle can be realized. Hereupon, the production of the
force to return this lever to the center can be realized
by two springs which are provided in the inside of a
switching device to give thereto a mechanical means such
as means of pulling the lever by uniform force from the
opposite sides, and so on.
Referring next to Fig. 28, ~here is a block
diagram showing the functional structure of a speech
speed conversion apparatus provided with an AV control
means. Referring to Fig. 29, there is a view for
explaining the operation of the AV control means in this
embodiment of Fig. 28. Referring to Figs. 30 and 31,
there are flow charts showing the operating procedure of
the main process in the speech speed conversion
apparatus provided with the AV control means in this
embodiment.
- 72
As shown in Fig. 28, the speech speed
conversion apparatus provided with the AV control means
in this embodiment is provided as a functional structure
in which an AV controller 28 is added to the functional
structure of the speech speed conversion apparatus in
the aforementioned embodiment shown in Fig. 11 and
connected to the controller 23E.
The aforementioned controller 23E judges
whether a condition for outputting an AV control signal
is satisfied or not and operates the AV controller 28 to
start/stop the outputting of the AV control signal.
As shown in Fig. 29, the AV control means is a
software in which the AV control signal is outputted
when the quantity of the lag from the real time as
caused by slow or repeat reproduction exceeds a pre-
determined value t30 seconds in Fig. 29) and in which
the outputting of the same signal is stopped when the
lag quantity then reaches zero via catching-up repro-
: ~:
duction. -
The AV control signal is picked out of this
apparatus and used for temporarily stopping the repro-
ducing operation of a recording/reproducing apparatus
such as a tape recorder, a video tape recorder, or the ~-~
like. By this means, it is made possible to continue
the slow hearing of an input voice which is continued
for such a long time that exceeds the capacity of the
ring buffer 24 in this apparatus.
In Figs. 30 and 31, the portion surrounded by
'' ,'-.`.':,',
"'~". .''`.
- 73 -
th~ broken line is a step showing the operating
procedure of the AV control means added to the flow
chart~ in Figs. 12 and 13. :[n this step, a judgment is
made as to whether the condition for outpu~ting the AV
contro~ signal is satisfied or not (S301). The judgment
with respect to the outputting of the AV control signal
is realized by a judgment as to whether the quantity of
the lag from the real time in a loop in which the one-
~rame waveform expansion/reduction process is repeated
for slow or repeat reproduction is over 30 seconds or
not (S301) and by starting the outputting of the AV
control signal when the lag from the real time is over
30 seconds (S302).
On the other hand, the process of stopping the :
AV control signal is carried out just after the judgment
"la~ quantity=0?" which is a judgment for escape out of
the loop of the catching-up reproduction process (S303).
Referring next to Figs. 32A, 32B and 32C,
there are views for explaining the arrangement of a
microphone in a speech speed conversion apparatus
according to the present invention. The reference
numeral 101 designates a body of the speech speed
conversion apparatus; 321, a microphone; 322, a prop
capable of expansion and contraction for supporting the
microphone 321; 323, a flexible prop for supporting the
microphone 321; and 324, an electric cord for electri-
cally co~necting the microphone 321 to the speech speed
conversion apparatus body 101 by wire.
- 74 -
Fig. 33 is a view showing a modified example
of this embodiment of Fig. 32, in which the reference
numeral 101 designates a body of the speech speed
conversion apparatus; 104 a slow switch; 105, a repeat
switch; 106, a reset switch; 321, a microphone; 324, an
electric cord for electrically connecting the microphone
321 to the speech speed conversion apparatus body 101;
325, an earphone; and 300, a connection member.
In the arrangement of the microphone in the
speech speed conversion apparatus of this embodiment, as
shown in Fig. 32A, the microphone 321 is supported by
. .: -.
the prop 322 capable of expansion and contraction.
Because the aforementioned supporting of the microphone
321 makes the microphone 321 far away from the speech
speed conversion apparatus body 101, the rustle of
clothes can be prevented from being produced when the ; ;
apparatus body is put into a breast pocket in use.
Alternatively, as shown in Fig. 32B, the
microphone 321 is supported by the flexible prop 323. ;~
Being supported by such a manner, the microphone 321 is
separated from the speech speed conversion apparatus
body 101, and can be bent in a desired direction.
!' ' Accordingly, the rustle of clothes can be prevented from
being produced when the apparatus body is put into a
breast pocket in use.
Further, as shown in Fig. 32B, the microphone
321 and the speech speed conversion apparatus body 101
are electrically connected to each other by wire (or
:; ~',',
"'`"'`~'~' '
wireless). The S/N ratio can be improved because the
microphone 321 and the speech speed conversion apparatus
body 101 are electrically connected to each other by
wire (or wireless) as described above so that the
microphone 321 is disposed near the listener indepen-
dently of the speech speed conversion apparatus body
101 .
Further, as shown in Fig. 33, the speech speed
conversion apparatus body 101 and the microphone 321 are
electrically connected to each other by the electric
cord through the earphone 325 and the connection member
300. Further, operation switches such as the slow
switch 104, the repeat switch 105, the reset switch 106,
and so on, are provided on the aforementioned connection
member 300. In this manner, not only the rustle of
clothes can be prevented from being produced when the
apparatus body is put into a breast pocket in use but
also both the SfN ratio and the handling property can be
improved.
Referring next to Fig. 34, there is a view for
explaining a lag time display means in a speech speed
conversion apparatus according to a further embodiment
of the present invention. The reference numeral 341
designates a display portion; and 342, a display screen.
As shown in Fig. 34, the lag time display
means in this embodiment displays how much the speech of
a speaker :i5 delayed from the real xpeech speed at the ~ -
time of the aforementioned slow/repeat reproductionO
76 -
For example, assuming that one human image represents
the time lag of 10 seconds in Fig. 34, then the time lag
from the current time is expressed in the number of
displayed human images. In this manner, the quantity of
time lag from the current time is recognized visually.
~ccordingly, both speaker and listener can adjust the
, .
speech speed conversion easi:Ly, so that this apparatus
can be used so as to be easy to handle.
The visual display of the time lag is ;~
realized, for example, by the provision of a liquid
crystal display in the front center of the speech speed
conversion apparatus body shown in E'ig. 6 and by the
display of the display screen as shown in Fig. 34 on the
liquid crystal display. Further, this display portion ;
15 is controlled by a "liquid crystal display driver" (not ~ ~
shown) connected to the controller 23E in Fig. 11. ~`
Because the quantity of the time lag to be
displayed is continuously managed by the lag quantity
, ;.
counter in the main process shown in Figs. 13 and 14,
the numerical value of this lag quantity counter can be
converted at the conversion rate of one to lQ seconds so
:~ " ` ` "'
that a corresponding number of human images can be ;-
indicated on the aforementioned display. This display~
ing operation is carried out by the controller 23E in " ~-
Fig. 11 through the aforementîoned display driver and
the timing of rewriting the display is sufficient as
long as the rewriting is performed whenever the proc- -
essing of one frame is completed. For exampler this
; ~
~
.'`.''''-"'''.
- 77 -
displaying process is carried out between the steps S137
and S138 in Fig. 14.
Referring to Fig. 35, there is a view for
explaining an electric source device in a speech speed
conversion apparatus accordi:ng to the present invention.
The reference numeral 1000 dlesignates an apparatus
portion concerning the speech speed conversion appara
tus; 1, a DSP; 5, an A/D converter; 6, a D/A converter;
9, an analog amplifier; 10, an analog amplifier; 1001,
an electric source; îO02, an electric power supply line;
and 1003, a changeover switch.
In the speech speed conversion apparatus in
this embodiment, as shown in the state transition view
of Fig. 15, a standby mode is provided besides the
through mode so that entry into the standby mode is made
automatically when the through mode is continued for a
predetermined time. That is, when either slow switch or
repeat switch is pushed tturned on), clock frequency is
heightened so that each process is carried out.
Further, in the through mode, the DSP 1
operates with fast clock but power is wasteful because
the speech speed conversion process or the like is not
executed. In the standby mode, therefore, the operating
clock for the DSP 1 is lowered so that only I/O of data
is performed to thereby reduce consumed electric power.
Further, only storage into the memory is performed. In
this manner, a voice memory function is realized.
Further, as shown in Fig. 35, at the time of
. .
-:: ., : , . . . ......................... ..
".,
- 78 -
an analog through mode, the changeov~r switch 1003 i~
connected to a contact side to cut off the electric
power supply line 1002 and a:Lso connected to a contact
side to connect the analog amplifiers 10 and 9 directly
so that electric power is not supplied to the DSP 1, the
A/D converter 5, the D/A con~erter 6 and peripheral
digital circuits. At this t:ime, the storage into the
memory is not performed. That is, I/O analog systems
are connected directly so as to be operated simply as an ~ `
10 analog amplifier. The aforementioned changeover switch ~`
is provided as a switch of three stages, namely, an ON
stage, an OFF stage and an ON-OFF intermediate stage as
:: :
shvwn in Fig. 35, so that the analog through mode is ~
provided. '~ ~,
As is obvious from the above explanation, in
accordance with this embodiment, a switch of three stage ;~
consisting of an ON stage, an OFF stage and an ON-OFF
inkermediate stage is formed so that the analog through ~;
mode is provided. Accordingly, not only reduction in ~ ~
20 electric power can be attained but also the range of use ~: -
of the electric source can be widened.
Referring to Fig. 36, an embodiment in which
; the speech speed conversion means according to the ~`
present invention is applied to a telephone will be
25 described. The reference numeral 2000 designates the -
speach speed conversion means according to the present
invention; 3000, a body of the telephone; 3001, a
.....
transceiver; and 3002, a telephone line. ` ;
.. ..
,;;,-~ ..,"
- 79 -
As shown in Fig. 36, the telephone in this
embodiment is formed by inserting the speech speed
conversion means 2000 according to the present invention
between the handset 3001 and the telephone body 3000.
The speech speed conversion means 2000 is, for example,
shaped like a mount on which the telephone body 3000 is
put.
Further, in the case of a cordless handset or
cordless child transceiver 3001, the speech speed
conversion means 200? is inserted between the trans-
ceiver 3001 and the telephone body 3000 by wireless.
Further, the speech speed conversion means
according to the present invention may be used as a
speech speed conversion means in a switching system so
that it can be operated at the user's request.
In the aforementioned structure, a voice can
be heard over the telephone slowly. Further, because
the voice is fed back as a through voice to the speaker
side as well as the voice can be heard slowly to the
listener so that the speaker can speak ordinarily at the
time of telephone conversation with the aged or the
like, there is no fear of hard speaking.
Further, it is unnecessary that any A/D means
is provided in the inside of the speech speed conversion
means as long as the speech speed conversion means is
provided as a digital circuit.
Referring to Fig. 37, description will be made
as to an embodiment in which the speech speed conversion
' ~''' ' ' ' ' ' ' ' ' ' ~' ' : ' ' ' ! ; . , ,
- 80 -
means according to the present invention is applied to a
premises broadcasting system. The reference numeral
2000 designates the speech speed conversion means; 321,
a microphone; 325,an earphone; 4003, an amplifier; and
4004, a speaker.
In the telephone in this embodiment, as shown
in Fig. 37, the speech speed conversion means 200
according to the present invention is inserted between
the microphone 321, the earphone 325 and the amplifier i~
. . ~ .
4003 for the speaker 4004.
In the aforementioned structure, the listener
can hear a voice at a suitable speech speed even in the
case where the speaking person does not control the
speech speed conversion operation. For example, even in
15 the case where the speaking person talks volubly at a `~`
high speech speed (impetuous speed) selfishly, the
listener can hear at a suitable speech speed.
Further, the listener can hear at a suitable
speech speed from a speaker even in the case where a
speaking person speaks slowly.
As is obvious from the above explanation, the
present invention can be applied to technical fields ~
requiring speech speed conversion, such as for example ~ ~-
hearing aids, learning of languages, abroad traveling,
music, and so on, besides telephones, telephone line
switching systems and premises broadcasting. -~
:, . . . .
For example, in goods for learning of lan- ;-
guages and for abroad traveling, the present invention
. -:
, .
- 81 -
can be applied to the following cases.
(1) A voice recorded is heard continuously and slowly.
(2) The expansion rate is cllanged in accordance with
the improvement in level.
(3) A portion which was hard to understand when heard
at an ordinary speed is heard repeatedly and slowly.
(4) After heard slowly, a voice is heard again at its
original speed.
(5) After slow repeat, pronounce is imitated.
(5~ An imitated voice is heard in comparison with its
original voice.
(7) A plurality of persons hear one source
simultaneously at their favorite speech speeds.
Further, in combination with a digital audio
apparatus such as a tape recorder, a CD recorder, an MD
recorder, and so on, it is unnecessary that any A~D
converter is provided in the speech speed conversion
apparatus as long as the audio apparatus has a digital
output.
Further, in the music purpose, the present
invention can be applied as long as changes are made as
follows.
The judgment based on the power of an expanded
frame is not carried out (because tempo is disordered).
The pitch extraction range is widened compared with
the case of a ~oice.
The waveform expansion process i5 carried out on
the basis of the pitch of a fixed length. In the case
- 82 -
of a voice thereof, the pitch is detected so that
processing is made on the basis of the detected pitch.
A foot switch is provided so that a converting
operation can be carried out by the foot switch. This
S makes it possible to control a music instrument while -;
playing.
Although the present invention has been
described specifically on the basis of the embodiments
thereof, it is a matter of course that the present
invention is not limited to the aforementioned embodi-
ments and that various changes may be made without
departing from the spirit thereof. -
In brief, effects obtained by typical embodi-
ments of the present invention disclosed in this
15 application are as follows. ~;
(1) Because the speech speed conversion apparatus can
be used not only for a voice one-sidedly given to the
listener such as a radio voice but also in the situation ;~
of conversation, a voice to be subjected to speech speed ;--
conversion can be selected by the listener without any
disturbance of listener's owni speech. ~;
Further, in hearing aids, foreign language -
learning machines, telephones, and so on, talker's voice -~ -
can be heard at a slow speech speed without any change
of the characteristic of the voice.
(2) Effect;ive use of the memory, a raw voice repeat
function, a voice memory function, a repeat voice speech
speed conversion function, a fast-hearing reproduction
- ~3 -
function, and so on, can be provided.
(3) Because means for changing the speech speed to a
value selec~ed by the speech speed selection switch is
provided, the speech speed oE a voice to be heard can be
selected by the listener's oT~n will.
(4) Because means for repeating a reproduced voice in a
period in which ~he repeat switch is turned on is
provided, the speech speed of khe repeat voice can be
converted.
(5) Because a catching-up means for catching up to a
position of stored information to be heard is provided
in the speech speed conversion apparatus, widening of
the range of application of the speech speed conversion
apparatus, reduction in operating time, improvement in
handling property, and so on, can be attained.
(6) Because at least one of the speech speed conversion
switch, the speech speed selection switch, the repeat
switch and the reset switch is provided in a peripheral
portion which is on a side surface of the speech speed
conversion apparatus and easy to handle, widening of the
range of application of the speech speed conversion
apparatus, reduction in operating time, improvement in
handling pxoperty, and so on, can be attained.
(7) Effic:iency in the speech speed conversion process
can be imp:roved.
(8) Because not only the determination of the waveform
expansion/reduction process and of the silent-part
elimination process in the speech speed conversion
- 8~ -
process is made on the basis of comparison between the
power of a frame and the threshold but also the thresh-
old is changed in accordance with the loudness of the
input voice, the speech speecl conversion process can be
carried out in accordance wit:h the environmental
condition in use. :.
(9) Because the microphone cloes not catch click noise
of switches, the reproduced voice can be heard
accurately.
(10) Because the switches have respective surface forms
which are different in tactility so as to be identified
without seeing, handling property can be improved.
~;1,'' , '
(11) Because means for preventing the rustle of clothes :
against the microphone is provided, the entrance of ~
15 noise can be reduced. .
(12) Because display means capable of visually indicat-
ing the quantity of a time lag from the current time is
: . .
provided in a predetermined position of the speech speed
conversion apparatus, reduction in operating time, . .
improvement in handling property, and so on, can be
attained.
(13) Because a ring buffer is used as a memory means so :-
that means for managing the lag time by a counter : :
indicating the time lag on the ring buffer is provided, ~ .
complex calculation of pointer addresses in the repea~
process, the catching-up process, and so on, can be
performed easily.
(14) Because a standby mode and an analog through mode
: . .
,j.,.,,.,., ;,.. ........
- 85 -
are provided besides the through mode, reduction in
consumed electric power can be attained.
(15) Because the electric source switch is provided in
the form of three stages consisting of an stage, an
stage and an OFF intermediate stage so that the analog
through mode is provided, not only reduction in electric
power can be attained but also the range of use of the
electric source can be widened.
(16) Because the aforementioned speech speed conversion
means is provided be~ween the handset of a telephone and
the apparatus body, a voice to be subjected to speech
speed conversion can be selected by the listener without
any disturbance of the listener's own speech.
(17) A talker's voice over the telephone can be heard at
a slow speech speed without any change of the
characteristic of the voice.
(18) Because the speech speed conversion means is
provided in a telephone line switching system, a voice
to be subjected to speech speed conversion can be
selected b~ the listener without any disturbance of the
listener's own speech.
'
I
