US4087632A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Speech recognition system
Assignee: AT&T Inc
•
Filing: 1976-11-26
•
Granted: 1978-05-02
•
Expires: 1996-11-26
Abstract
Abstract A speech recognition system is realized by applying a speech signal to a feature extractor wherein a sequence of predetermined features representing the speech signal are determined, and by comparing the determined features, in an acceptor, to predetermined sequences of features which represent selected words. One attribute of the feature extractor is the ability to represent certain classes of sounds in terms of the position and direction of motion of the movable structures of a human vocal tract model, such as the position and direction of movement of the speaker's tongue body. The tongue body position is derived by determining the formant frequencies in the applied speech signal and by employing the Coker vocal tract model to find the tongue body position which best matches the determined formants.
Inventors
- 👤 Edward Henry Hafer
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-27 13:01
CPC Classifications
G; G10; G10L; G10L15/00
Claims Analysis
Total Claims: 16
Claim 1
Independent Claim
70 words
A speech recognition system responsive to sampled representations of applied speech comprising: an extended feature extractor, responsive to said sampled representations, for determining features of speech contained in said sampled representations, including a feature corresponding to the tongue body position and its direction of movement; and an acceptor, responsive to said feature extractor, for matching the sequence of said determined features to predetermined sequences of features corresponding to selected words.
Claim 2
Dependent Claim
70 words
A speech recognition system responsive to sampled representations of applied speech comprising: an extended feature extractor, responsive to said sampled representations, for determining features of speech contained in said sampled representations, including a feature corresponding to the tongue body position and its direction of movement; and an acceptor, responsive to said feature extractor, for matching the sequence of said determined features to predetermined sequences of features corresponding to selected words.
Claim 3
Dependent Claim
30 words
The system of claim 1 wherein the said features determined by said feature extractor comprise a Silence token, a Burst token, a Fricative token, and a Tongue body trajectory token.
Claim 4
Dependent Claim
76 words
The system of claim 1 wherein said feature extractor comprises: first means responsive to said sampled representation for computing Silence features; second means responsive to said sampled representations for computing Burst features; third means responsive to said sampled representations for computing Fricative features; fourth means responsive to said sampled representations for computing formant frequencies of said applied speech; and fifth means responsive to said fourth means for converting said format frequencies to Tongue body trajectory features.
Claim 5
Dependent Claim
25 words
The apparatus of claim 3 wherein said fifth means employs a vocal tract model to convert said formant frequencies to said Tongue body trajectory features.
Claim 6
Dependent Claim
14 words
The apparatus of claim 3 wherein said fifth means comprises a memory look-up table.
Claim 7
Dependent Claim
62 words
A speech recognition system having means for developing a sequence of features representing applied signals and having means for matching the succession of features to predetermined sequences of features representative of preselected words Characterized in that said means for developing a sequence of preselected features determines the formants contained in said applied signals and converts said formants to Tongue body trajectory features.
Claim 8
Dependent Claim
62 words
A speech recognition system having means for developing a sequence of features representing applied signals and having means for matching the succession of features to predetermined sequences of features representative of preselected words Characterized in that said means for developing a sequence of preselected features determines the formants contained in said applied signals and converts said formants to Tongue body trajectory features.
Claim 9
Dependent Claim
26 words
The system of claim 6 wherein said means for extracting converts said formants to said Tongue body trajectory features in accordance with Coker's vocal tract model.
Claim 10
Independent Claim
79 words
A method of recognizing speech by converting representations of intervals of said speech into sequences of preselected types of parameters including the steps of (1) identifying the formant frequency content of said speech during vowel intervals; (2) transforming said formant frequency content into tongue body position and direction of movement parameters; and (3) detecting a correspondence between said sequences of preselected types of parameters including said tongue body parameters with a preselected sequence of said preselected types of parameters.
Claim 11
Independent Claim
79 words
A method of recognizing speech by converting representations of intervals of said speech into sequences of preselected types of parameters including the steps of (1) identifying the formant frequency content of said speech during vowel intervals; (2) transforming said formant frequency content into tongue body position and direction of movement parameters; and (3) detecting a correspondence between said sequences of preselected types of parameters including said tongue body parameters with a preselected sequence of said preselected types of parameters.
Claim 12
Independent Claim
110 words
A method for recognizing connected speech comprising the steps of analyzing an applied speech signal in search for silence intervals and developing a silence token whenever such intervals are present; analyzing said speech signal in search for burst intervals and developing a burst token whenever such intervals are present; analyzing said speech signal in search for fricative sound vowels and developing a fricative token whenever such intervals are present; analyzing said speech signal in search for vowel sound intervals and developing a tongue body trajectory token whenever such intervals are present; and comparing the succession of developed tokens to preselected sequences of tokens characteristic of prechosen words to be detected.
Claim 13
Independent Claim
110 words
A method for recognizing connected speech comprising the steps of analyzing an applied speech signal in search for silence intervals and developing a silence token whenever such intervals are present; analyzing said speech signal in search for burst intervals and developing a burst token whenever such intervals are present; analyzing said speech signal in search for fricative sound vowels and developing a fricative token whenever such intervals are present; analyzing said speech signal in search for vowel sound intervals and developing a tongue body trajectory token whenever such intervals are present; and comparing the succession of developed tokens to preselected sequences of tokens characteristic of prechosen words to be detected.
Claim 14
Dependent Claim
38 words
The method of claim 9 wherein said step for developing tongue body trajectory tokens comprises the steps of: analyzing said speech signal to develop the formant frequency contained therein; and correlating said developed formants to tongue body positions.
Claim 15
Independent Claim
66 words
A method for recognizing speech uttered by a speaker by extracting from said speech a sequence of tokens representative of said speech, said tokens including a class of tokens characterizing said speaker's tongue body trajectory, comprising the steps of: deriving the values of the formant frequencies contained with said speech; and converting said derived formants to tongue body positions in accordance with a vocal tract model.
Claim 16
Independent Claim
66 words
A method for recognizing speech uttered by a speaker by extracting from said speech a sequence of tokens representative of said speech, said tokens including a class of tokens characterizing said speaker's tongue body trajectory, comprising the steps of: deriving the values of the formant frequencies contained with said speech; and converting said derived formants to tongue body positions in accordance with a vocal tract model.
References Cited (Backward Citations)
US3129287A
Cited
US3395249A
Cited
US3530248A
Cited
US3624302A
Cited
US3679830A
Cited
US3755627A
Cited
US3723667A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1981-05-01 | 1981-10-31 | 1982-05-02 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1985-05-01 | 1985-10-31 | 1986-05-02 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1989-05-01 | 1989-10-31 | 1990-05-02 | $7700.00 | Window Closed / Lapsed |