Patent US4811399A successfully refreshed from Google Patents.
US4811399A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Apparatus and method for automatic speech recognition
Assignee: ITT Inc
•
Filing: 1988-06-14
•
Granted: 1989-03-07
•
Expires: 2008-06-14
Abstract
Abstract An apparatus for automatic speech recognition includes a plurality of processors, including template processors in which there are stored templates representative of both speech and non-speech sounds. Incoming sounds are continuously converted into digital signals in respective frames representative of speech and non-speech sounds, respectively. Sequences of such frames are compared with both the speech and non-speech templates to determine the closest matches. Endpoints of respective speech utterances are determined in response to the detection of respective non-speech-speech-non-speech sequences, whereupon such speech utterances are processed to recognize the same.
Inventors
- 👤 Blakely P. Landell
- 👤 Robert E. Wohlford
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-30 03:26
CPC Classifications
G; G10; G10L; G10L25/00; G10L25/78; G10L25/87; G10L15/00; G10L15/08; G10L15/10; G10L15/12; G10L25/03; G10L25/18
Claims Analysis
Total Claims: 5
Claim 1
Independent Claim
140 words
An apparatus for recognizing speech sounds contained in audio speech input comprised of sound utterances separated by silences of at least a minimum silence duration, wherein said sound utterances may include non-speech sounds other than ambient noise, comprising: frame forming means for continuously forming frames of digital signals representative of said speech sounds, silences, and said non-speech sounds; recognizing means for continuously recognizing said speech sounds, silences and said non-speech sounds from said frames of digital signals; determining means for determining endpoints of said speech sounds based on the recognition of non-speech-speech-non-speech sequences, wherein the non-speech portions of the respective sequences include recognized silences as well as recognized non-speech sounds other than the ambient noise; and processing means for processing at least the speech portions of the respective sequences demarcated by the determined endpoints in accordance with a predefined syntax.
Claim 2
Independent Claim
140 words
An apparatus for recognizing speech sounds contained in audio speech input comprised of sound utterances separated by silences of at least a minimum silence duration, wherein said sound utterances may include non-speech sounds other than ambient noise, comprising: frame forming means for continuously forming frames of digital signals representative of said speech sounds, silences, and said non-speech sounds; recognizing means for continuously recognizing said speech sounds, silences and said non-speech sounds from said frames of digital signals; determining means for determining endpoints of said speech sounds based on the recognition of non-speech-speech-non-speech sequences, wherein the non-speech portions of the respective sequences include recognized silences as well as recognized non-speech sounds other than the ambient noise; and processing means for processing at least the speech portions of the respective sequences demarcated by the determined endpoints in accordance with a predefined syntax.
Claim 3
Dependent Claim
107 words
The apparatus of claim 1, wherein said recognizing means includes memory means for storing representations of speech and non-speech sounds as respective speech and non-speech templates, wherein said non-speech templates include a silence template of the minimum silence duration, an ambient noise template for the ambient noise, and multiple non-speech noise templates for noises other than the ambient noise having a duration several times longer than the minimum silence duration; and wherein said processing means includes means responsive to an endpoint determination by said determining means for recognizing said speech and non-speech sounds by comparing the respective frames of original signals with said speech and non-speech templates.
Claim 4
Independent Claim
177 words
A method of recognizing speech sounds contained in audio speech input comprised of sound utterances separated by silences of at least a minimum silence duration, wherein said sound utterances may include non-speech sounds other than ambient noise, comprising the steps of: storing representations of speech sounds and non-speech sounds as respective speech and non-speech templates in a memory, wherein said non-speech templates include a silence template of the minimum silence duration, an ambient noise template for the ambient noise, and multiple non-speech noise templates for noises other than the ambient noise; continuously converting said speech sounds, silences and said non-speech sounds into frames of digital signals; determining endpoints of said speech sounds from respective non-speech-speech-non-speech sequences, including continuously comparing said frames of said speech sounds, silences, and non-speech sounds with said speech and non-speech templates until detection of the respective sequence and processing at least those of said frames in the respective sequence that are representative of speech sounds which are demarcated by the determined endpoints in accordance with a predefined syntax to recognize such speech sounds.
Claim 5
Independent Claim
177 words
A method of recognizing speech sounds contained in audio speech input comprised of sound utterances separated by silences of at least a minimum silence duration, wherein said sound utterances may include non-speech sounds other than ambient noise, comprising the steps of: storing representations of speech sounds and non-speech sounds as respective speech and non-speech templates in a memory, wherein said non-speech templates include a silence template of the minimum silence duration, an ambient noise template for the ambient noise, and multiple non-speech noise templates for noises other than the ambient noise; continuously converting said speech sounds, silences and said non-speech sounds into frames of digital signals; determining endpoints of said speech sounds from respective non-speech-speech-non-speech sequences, including continuously comparing said frames of said speech sounds, silences, and non-speech sounds with said speech and non-speech templates until detection of the respective sequence and processing at least those of said frames in the respective sequence that are representative of speech sounds which are demarcated by the determined endpoints in accordance with a predefined syntax to recognize such speech sounds.
References Cited (Backward Citations)
US4481593A
Cited
US4489435A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1992-03-06 | 1992-09-05 | 1993-03-07 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1996-03-06 | 1996-09-05 | 1997-03-07 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2000-03-06 | 2000-09-05 | 2001-03-07 | $7700.00 | Window Closed / Lapsed |