Note: Descriptions are shown in the official language in which they were submitted.
CA 02205550 l997-05-20
WO 97/12361 PCT/US96/15283
TELEPHONE NEIWORK SERVICE FOR
CONVERTING SPEECH TO TOUCH-TONES
Field of the Invention
The present invention relates generally to the field of telecommunications and
5 more particularly to a method for providing a speech-based interface to DTMF (Dual
Tone Multi-frequency) controlled systems.
R~ r ' of the Invention
In recent years, the teleco,~ ui-ications industry has seen an explosive growth
in the use of DTMF (i.e., touch-tone) controlled systems. These systems have typically
io been employed either for the purpose of providing added user convenience or
~-cescihility to information, or to reduce or elimin~t~ the need for human telephone
operators. For example, voice m~c~ging systems often may be ~rcesse~ with use oftouch-tones to enable a caller to retrieve his or her messages or to record outgoing
greeting mloss~ges or pelr~ , other ~-lminictrative tasks. Even home-based telephone
15 ansv~el;l.g ~ ;n~s now routinely provide remote access from any telephone through
touch-tone control. In addition, many customer service telephone numbers presenttouch-tone controlled menus to the caller, where the menus can be used to efficiently
provide a connection to the a~lo~,iate person or dep~~ .l. Moreover, touch-tonesare used to control many other ~ ul..~l~ci systems, such as, for example, au~u~llat~d
20 banking transaction and stock quotation services. These services often enable a caller
to pelr~,ll an entire transaction without involving a human participant on the other side
of the call.
Although the fraction of telephones without touch-tone capability (i.e., rotary or
dial pulse type telephones) is small and declining in the United States, there are still a
25 substantial number of such telephones, especially outside of the United States.
Moreover, many modern touch-tone telephones -- including, for example, most
cordless and cellular telephones -- provide the keypad in the h~n-lset, thereby making
CA 02205550 l997-05-20
WO 97/12361 PCT/US96/1~283
it extremely cumbersome for a user to provide the required control signals to such an
automated touch-tone based system while co~ g to listen to the responses from the
system. Thus, the use of these automated systems with typical cordless or cellular
telephones often involves frequent back and forth movelnellt of the handset which can
5 become quite annoying. In addition, many telephone users, even those who are not
facing any of the above-mentioned limitations, would find it more convenient if they
could merely speak a comm~n-~, rather than having to locate and then press an
ap~rol)~iate key with their finger.
Meanwhile, speech recognition technology has advanced substantially over the
10 last several years. It is now technologically strai~hLru. w~d to provide speaker
independent speech recognition (i.e., recognition of speech from any speaker previously
unknown to the system) on individual words selected from a modest vocabulary. Infact, some of the afore.~ .t;on~A ~ o...~t~d systems provide such speech recognition
capability, eli...;.~ the need for the user to press touch-tone keys. On the other
hand, it may still be quite a while before the vast lllajo~ily of these systems will have
incol~.,.ated speech recognition capability to replace (or su~pl~ ,.ll) their present touch-
tone control mPe~l~nieme, since, for example, the investment required to add speech
recognition capability to such a system may not be justifiable on the basis of a single
individual system alone.
It would be advantageous, thel~erole, to provide a generalized capability enabling
the user of a telephone to speak a word .~ se~lt;.lg a digit (or letter) key, the "*" key
(e.g., the word "star") or the "#" key (e.g., the word "pound"), and to have theco..~;s~ ding DTMF signal (i.e., touch-tone) ~Ul~ ;r:~lly genelatt~d in order to provide
the lc~luin,d input to an ~IJiL ~ touch-tone controlled system. U. S. Patent No.25 5,402,477, issued to Michael L. McMahan and Michele B. t'~mmf~.l on March 28, 1995,
discloses a telephone set which, inter alia, includes the capability of gelle,atillg touch-
tone signals in response to â digit, the word "star," or the word "pound," orally spoken
by the user into the h~nrl~et But to make use of this capability, it is obviously
n~ces~ry that this particular specialized telephone terminal e~ .,P~-t as disclosed
-
CA 0220~0 1997-0~-20
Wo 97/12361 PCT/US96/15283
therein be available to the user. As such, the approach of U. S. Patent No. 5,402,477
clearly does not provide a solution to the problem described above and addressed herein
-- namely, that many telephone calls are, in fact, ~;ullcllLly being placed either with
rotary or pulse type telephones or with touch-tone telephones whose keypads are located
5 in the handset, and will contin~le to be so for the foreseeable future.
S~ r~ of the Invention
The present invention provides a telecollllllullications network service for
converting spoken words to individual DTMF (e.g., touch-tone) signals to be furnished
to an ~llt{lm~te~l system responsive to touch-tone control thereof. In accordance with
10 one illustrative embodiment of the invention, a telephone user speaks an uLL~,Ia~ce
comprising a word lc~lc~e~ g a digit, letter, the "*" key (e.g., the word "star") or the
"#" key (e.g., the word "pound"). The illustrative network service performe
conventional, speaker independent speech recognition on the uLL~.~lce to identify the
spoken word and then gc.lcldtes the touch-tone signal which corresponds to the spoken
1~ word so identified (Gg., a digit, letter or special character as found on a conventional
touch-tone telephone). This cl.~,.ated DTMF signal is then ~ cl to an aululllated
system such as a voice m~ee~ing system or telephone allswcling m~t hinP to control the
operation thereof. The ll~,.W~lk service of the present invention may be advantageously
employed when the telephone being used by the caller is a rotary or dial pulse type
telephone, or in other cir.;.. ~ es where it may be preferable to speak rather than to
press the keys of a telephone keypad.
Brief DescriPtion of the Drawin~s
Fig. 1 shows an illu~ ive teleco.~ tion~ network in which a speech to
touch-tone co"~ ion network service has been provided in accordance with an
25 illustrative embodiment of the present invention.
Fig. 2 shows a flow chart describing the operation of the illustrative speech totouch-tone conversion server of Fig. 1 in accordance with an illustrative embodiment of
CA 02205550 1997-05-20
WO 97/12361 PCT/US96/1~;283
the present invention.
Detailed DescriPtion
Fig. 1 shows an illustrative teleco~ c~tions network in which a speech to
touch-tone conversion network service has been provided in accordance with an
5 illustrative embodiment of the present invention. A caller uses telephone 11 in a
conventional ll~almcl to place a telephone call to a particular telephone uulllb~,r which
has been ~c~ign~d to speech to touch-tone con~ ion server 15. Telephone 11 may be
any type of co.~ .lional telephone, although the use of the present invention will be
found to be particularly advantageous if telephone 11 is a rotary or dial pulse type
10 telephone, or, ~lterT~tively, a touch-tone telephone such as a cordless or cellular
telephone wherein the touch-tone keypad is located in the h~n-l~et In response to the
placement of the call by telcphone 11, teleco"-",--.,;c~tions network switch 13
interconncc~ telephore 11 with speech to touch-tone con~ ion server 15.
Teleco"...~ ir~tions network switch 13 may be a conventional long ~ fe or local
exchange tcleco-.. ~.. ir~tions switch, f~mili~r to those of ordinary skill in the art.
Once the call ~~ telephone 11 and speech to touch-tone conversion server
15 has been coln-~ the server l~que~l~ that the user provide the telephone number
of an au~o~lal~,d system l.,~onsi~,., to touch-tone control thereof, such as a voice mail
system, a telepho~e allS~ illg ~ rk;~ or an aulolllated transaction service such as a
20 banking or stock quotation service. The requested telephone ..u...bcr may be provided
in a c.,n~e~l;nn~ ulnc. tie.~ by dialing), or it may be provided orally by the caller.
In the latter case, co.l~,..lional speech recognition techniques may be used to determine
the se~ re of digits which compn~es the telephone IlUlllbCr of the desired au~ ated
system. In the illustrative embodiment of Fig. 1, the user specified telephon.o number
25 is that of ~ulo~llated touch-tone responsive system 19.
Once the server has been provided with the telephone number to which the caller
wishes to be connecled, it may use any of a l-u---bel of convc--Lional techniques known
to those skilled in the art to effectuate a colme-;lion bet~..,en the caller and automated
CA 02205550 1997-05-20
WO 97/12361 PCT/[JS96/1~;283
touch-tone responsive system 19, with the server itself interposed therebetween. In the
illustrative embodiment of Fig. 1, for example, such a connection is established via
teleco"""l~,ir~tions network switch 17, which, like teleco""..~ tions network switch
13, may be a conventional long ~ t~nre or local exchange telecommllnic~tions switch.
5 (Note that network switch 17 may, in fact, be the same switch as network switch 13.)
In one embodiment, for example, the functionality of speech to touch-tone
conversion server 15 may be provided with use of a PBX (private branch exchange)system which is directly connectecl to network switch 17. This connection between the
PBX and the network switch may, for exarnple, comprise a set of Tl trunk lines.
10 familiar to those of oldi.l~y skill in the art, and may use conventional digital .sign~lling
techniques to co~ ni~tç therebetween. In particular, speech to touch-tone conversion
server 15 may be connected on the line side of the afG~ elltioned PBX as if it were
one or more telephone sets. (Note that speech to touch-tone conversion server 15 may,
in general, comprise a plurality of ports, thereby enabling its .sim~lt~n~ous use by a
15 llunlb~,. of individual callers, each caller co""..u..i~ting with an individually associated
automated touch-tone l~p~,lsive system.)
In use, the call from tclephone 11 to server 15 may be set up through the
aforementioned PBX. In this manner, the caller will be col~nec~d to a first line of the
server. Then, as described above, speech to touch-tone conv._~ion server 15 requests
20 from the caller the telephone IlUl~ of automated touch-tone responsive system 19.
Once the server has received that llulllber, it al~cesses a second line (to the PBX) and
requests that the PBX system dial the call to the ~Luto..,ated system, c-,..necli..g the
server thereto. The server then instructs the PBX to conference the first line and the
second line together, thereby reSllltinE in the caller being co~ ed to the ~uLolllated
25 system with the server interposed therebetween. At this point, the server is poised to
provide speech to touch-tone conversion in accordance with an illustrative embodiment
of the present invention as described below.
Fig. 2 shows a flow chart describing the operation of the illustrative speech totouch-tone conversion server of Fig. 1 in accordance with an illustrative embodiment of
CA 0220~0 1997-0~-20
Wo 97~12361 PCT/USg6/15283
the present invention. Specifically, steps 21-25 yelrullll the process of interconnecting
the caller with the user specified telephone number as described above, with the server
interposed between the caller and the called number. In step 2i, speech to touch-tone
conversion server 15 answers the incoming call (on what will be referred to herein as
5 the "first" line). Step 22 then instructs the caller to provide the telephone number of
automated touch-tone responsive system 19 -- the automated system to which the caller
wishes to be connl~ct~i In step 23, the server receives the telephone number from the
caller. As described above, this may, for example, colll~lise the conventional
recognition of speech utterances which specify the llulllber, or it may comprise other
10 conventional techniques such as the receipt of incorning touchtones.
Once speech to touch-tone conv~l~ion server 15 has received the telephone
number to be called, it requests a second (outgoing) line from the telephone network,
and dials thereon the user-provided number of the auLolllaled touch-tone responsive
system (step 24). And, finally, when the connection on the second line has been
15 completed, the server causes the two lines (i.e., the first line which includes the caller
and the second line which includes the auLo.llated system) to be "confe.~,need" together
(i.e., illt.,l~o~ eclt-d) in step 25. This may be acconlplished either intern~lly within the
server, or the server may instruct the telephone network to merge the two lines into a
single illl.,..;onn~ d line. In any event, at this point, the server is poised to provide
20 speech to touch-tone convelsion in acco~ance with an illu~halive embo~im~nt of the
present invention as shown in steps 31-34.
Specifically, steps 31 through 34 provide speech to touch-tone conversion as
needed until the server is disconn~cte~i from the call by the caller (or until the call itself
beculll~s discon..~ l by either party) as d~ llilled by decision 35. In particular, step
25 31 listens for a spoken word from the user. Note that for many automated touch-tone
responsive systems, the only speech the caller is likely to produce is that provided in
accordance with the present invention for the pul~ose of genelating touch-tones. Thus,
in such a case, it may be advantageously p.esu-l-~d that the spoken word is one which
represents either the digits "0" through "9, "the letters "A" through "Z," or one of the
CA 02205550 1997-05-20
WO 97112361 PCT/US96/15283
special character found on conventional touch-tone telephones (i.e., "*" or "#").
However, for systems which ultimatel-/ may connect the caller to another person, for
exarnple, the caller may advantageously be provided with a mechanism for disconnecting
the server from the call before the call itself has been disconnecte~l thereby elimin~ting
5 the possibility that the server will erroneously aKempt to generate DTMF tones in
response to conversational speech by the caller.
As the word for each digit or letter is uttered, speech to touch-tone conversionserver 15 pe.folll.s conventional, speaker independent speech recognition on the spoken
utterance in step 32, thereby identifying the spoken word as one of the words
representing the digits "0" through "9," the letters "A" through "Z," or one of the special
characters "*" or "#." In particular, the set of words used to represent the digits may,
for example, include the words "one," "two," "three," "four," "five," "six," "seven,"
"eight," "nine," "zero" and "oh." Similarly, the set of words used to l~,plc:sellt the
character "*" may include the words "star" and ~cteri~ and the set of words used to
lcplcsent the character "#" may include the words "pound," "sharp" and "number."Moreover, words in languages other than F.nglieh may obviously be provided either in
addition to or instead of the f~mili~r Fn~lieh words for the digits, letters and special
characters. Based on the above-described speech r~co~l.ilion process, step 33 cl~csifiçs
the itlentifie.l word into the ~proy~iate tclepl1ol~e push button category.
Once the spoken word has been irl~ntified and cl~ceifi~ the server then employs
a DTMF signal generator to gcn~aLc thc touch-tone signal which collcs~onds to the
given telephone push button category (step 34), thereby providing the n~cecs~.y input
to the automated system. For ex~nplP if the iflPntifie~l word is either "two," "A," "B,"
or "C," the touch-tone signal which has been ~csign~ to the digit "2" for use in touch-
tone based telcco~ r~ti~nc si n~llin~ will be ~pr~liately ge.~e~dted. (As is well
known to those skilled in the art, DTMF or touch-tone signals comprise a matrix
combination of two frequenr,i~c, each from a group of four, which are used in the
operation of push button telephone sets to transmit numerical address information.
Specifically, the two groups of four frequencies have been standardized to be 697 Hz,
CA 0220~0 1997-0~-20
Wo 97/12361 PCT/US96/15283
770 Hz, ~52 Hz, and 941 Hz, and 1209 Hz, 1336 Hz, 1477 Hz, and 1633 Hz.)
As pointed out above, the server provides speech to touch-tone conversion
continuously as needed until the caller requests that the server be removed from the call,
at which point the server disconnects itself from the line (i.e., both the first line and the
5 second line) in step 36. (The server will also advantageously disconnect itself from the
line when the caller hangs up.) Such a request can be communicated to the server in
a number of possible ways. For example, the word '~discolme~l can be added to the
vocabulary of the server's speech recognition capability, and upon recognizing that the
word "disconnect" has been spoken by the caller, the server disconnects itself from the
lO call. Many other possible mech~nicmc for terrnin~ting the server's participation in the
call will be obvious to those skilled in the art.
Although a specific embodiment of this invention has been shown and described
herein, it is to be understood that this embodiment is merely illustrative of the many
possible specific arran~;c~ which can be devised in application of the principles of
15 the invention. Numerous and varied other arrangements can be devised in accordance
with these principles by those of ordinary skill in the art without departing from the
spirit and scope of the invention. For example, l1W11~OUS alternative m.oc~l~ni.cmc may
be employed for hltelco~ e~ g a calling party with an automated system responsive
to touch-tone control, with a speech to touch-tone conversion capability interposed
20 therebetween, at which point the prin~ip~l~ of the present invention may be used to
convert spoken words to touch-tones for use by the automated system. In addition,
some or all of the functions of speech to touch-tone conversion server 15 as described
in the illustradve embodiment shown herein may, for example, be incol~o~dlcd directly
into one or more telecu~ ic~tions network switches (such as ~wilchcs 13 and/or 17
25 of Fig. 1), or they may be provided within the telecol-~ --ications network by any of
a number of other possible mPr~nicm~ obvious to those skilled in the art. Moreover,
as pointed out above, speech to touch-tone conversion server 15 may be provided with
the capability of simultaneously h~n-lling a plurality of independent telephone calls,
thereby providing a generally available network service.
