Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
1999-12-02
2001-04-17
Hudspeth, David (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S231000, C381S091000
Reexamination Certificate
active
06219645
ABSTRACT:
RELATED APPLICATIONS
None.
FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
Not Applicable.
BACKGROUND OF THE INVENTION
This invention relates to voice control of electronic devices, control of a number of distributed processing systems, multi-pickup sensing apparatus, and to triangulation of a user's position.
Voice control of various electronic devices and networks is becoming more prevalent. Advances in signal processing hardware and algorithms are making it possible to build automatic speech recognition (ASR) voice response systems with larger vocabularies and higher accuracy. Also, inexpensive, highly integrated microelectronic circuitry is enabling the inclusion of simple voice response systems in many classes of low cost consumer products, including appliances, entertainment systems, voice-dialing telephones such as the GENIE(™) GSM cellular telephone by Phillips Electronics in the Netherlands, and toys such as FURBY(™) by Tiger Electronics of China Manufacturers of such microelectronic circuitry include Lucent, Motorola, Texas Instruments, and others, and general purpose microprocessor manufacturers like Intel, Motorola, IBM, and AMD.
Today's ASR-based voice response systems all use a single signal input that is picked up by a single microphone, processed by a signal processor or other processing means to extract the speech features, and sent through a recognition algorithm that matches the input signals with stored templates or models. The controlled device then uses the output of the match algorithm to control some aspect of its operation. Examples of today's voice response systems include interactive computer games and dictation software packages available for personal computers at any computer retailer, and products available from Lucent, such as Service Circuit System and Compact Service Node.
Several limitations exist with current implementations of ASR voice response systems. Current implementations are very susceptible to errors caused by ambient sound noise interfering with the speech. The microphone picks up ambient sounds in addition to the voice signal of interest. The voice recognition algorithms use voice models that assume a pure voice sound. Mixing ambient noise with the desired pure signal leads to reduced accuracy of the algorithm.
Special microphones are usually mounted quite close to the user's mouth on head-mounted booms are often employed to lessen ambient noise effects. Head mounted booms are sold by specialty telephone retailers, computer stores, microphone manufacturers including Shure Brothers, and are available at mass market retailers including RADIO SHACK(™) stores, a division of Tandy Corp., in Fort Worth, Tex. Current art uses many techniques, such as moving the microphone closer to the source (as a boom microphone does), using highly directional microphones, and using soundproof rooms to contain the speaking.
Another limitation of the present art is that only a single ASR device can be used at a time. If a user has several ASR devices in use simultaneously, for example a computer, telephone, and intelligent home controller, it may be necessary to switch microphones often in order to talk to and therefore control the desired ASR device.
Since most head mounted microphones currently in use for ASR are corded (again to reduce interference and noise pickup), the user often feels tethered to the electronics box, hampering mobility through a room on environment.
SUMMARY OF THE INVENTION
A solution to all of the problems in the current art involves using an array of microphones spread throughout the working volume of a room. All of these microphones are fixed in location to the walls and furniture. Their signals are all directed to a signal processing circuit that compares the amplitude, frequency content, and phase of the signals picked up from the various microphones, and decides the location of the user, and also estimates the user's head orientation.
Diversity noise cancellation techniques improve the signal to noise ratio of the post-processed signal by choosing which microphone(s) are providing the highest quality audio before passing it on to the ASR algorithm.
A new control mode is possible using a diversity microphone configuration, wherein the user is surrounded by several devices that can accept voice inputs. Using the present invention, the user would turn directly to face the device to be controlled, and speak the command. Only the device that is in alignment with the direction the user was facing would respond to the voice command.
Advantageously, the directional ASR system reduces the number of false command executions caused by the user saying command keywords in normal speech, because voice commands would only execute to completion if the user were facing a particular device to within a small, programmable angle tolerance.
Also advantageously, the system of the present invention greatly improves the performance and accuracy of voice control by reducing the size of the recognition vocabulary required by the ASR algorithms.
These and other aspects and attributes of the present invention will be discussed with reference to the following drawings and accompanying specification.
REFERENCES:
patent: 5561737 (1996-10-01), Bowen
patent: 5583965 (1996-12-01), Douma et al.
patent: 5657425 (1997-08-01), Johnson
patent: 5774841 (1998-06-01), Salazar et al.
patent: 5828997 (1998-10-01), Durlach et al.
patent: 6014626 (2000-01-01), Cohen
patent: 6052666 (2000-04-01), Diehl et al.
Web Site printout, Georg Neumann GmbH, Berlin, KLU 100 Dummy Head microphone (15 pgs).
Web Site printout, Sensory, Inc., Sunnyvale, CA, (2 pgs).
Web Site printout, Wizzard Software, Pittsburgh, PA (3 pgs).
Web Site printout, IBM Software, “ViaVoice”(2 pgs).
Web Site printout, Philips Electronics, “PSP—Professional Solutions—Voice Control” (2 pgs).
Texas Instruments,TMS320 DSP Produce Family—Glossary, Feb. 1998, pp. I-117.
Web Site printout, Analog Devices Product Page: ADSP-21065L; Products and Datasheets (4 pgs).
Web Site printout, Texas Instruments, “DSP Overview at Texas Instruments” (2 pgs).
Web Site printout, Analog Devices Product p.: SSM2250; Product p.: AD 1877; Product p. 1879; Products and Datasheets (8 pgs).
Web Site printout, Telex Computer “Audio—Microphones & Headsets for speech dictation” (3 pgs).
Web Site printout, Shure Brothers Inc. “Shure SM89 Shotgun Condenser Microphone” (5 pgs).
Analog Devices Data Sheet, DSP Microcomputers, ADSP-21msp58/59, 1995, pp. 1-40.
Azad Abul K.
Hudspeth David
Lucent Technologies - Inc.
Sitrick & Sitrick
LandOfFree
Enhanced automatic speech recognition using multiple... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Enhanced automatic speech recognition using multiple..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Enhanced automatic speech recognition using multiple... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2517788