US5799276A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Knowledge-based speech recognition system and methods having frame length computed based upon estimated pitch period of vocalic intervals
Assignee: Lexia Learning Systems Inc
•
Filing: 1995-11-07
•
Granted: 1998-08-25
•
Expires: 2015-11-07
Abstract
Abstract Knowledge based speech recognition apparatus and methods are provided for translating an input speech signal to text. The speech recognition apparatus captures an input speech signal, segments it based on the detection of pitch period, and generates a series of hypothesized acoustic feature vectors for the input speech signal that characterizes the signal in terms of primary acoustic events, detectable vowel sounds and other acoustic features. The apparatus and methods employ a largely speaker-independent dictionary based upon the application of phonological and phonetic/acoustic rules to generate acoustic event transcriptions against which the series of hypothesized acoustic feature vectors are compared to select word choices. Local and global syntactic analysis of the word choices is provided to enhance the recognition capability of the methods and apparatus.
Inventors
- 👤 Edward Komissarchik
- 👤 Vladimir Arlazarov
- 👤 Dimitri Bogdanov
- 👤 Yuri Finkelstein
- 👤 Andrey Ivanov
- 👤 Jacob Kaminsky
- 👤 Julia Komissarchik
- 👤 Olga Krivnova
- 👤 Mikhail Kronrod
- 👤 Mikhail Malkovsky
- 👤 Maxim Paklin
- 👤 Alexander Rozanov
- 👤 Vladimir Segal
- 👤 Nina Zinovieva
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-28 01:49
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/08; G10L15/18; G10L15/04; G10L15/14; G10L15/142; G10L15/183; G10L15/187; G10L15/19; G10L15/193
Claims Analysis
Total Claims: 58
Claim 1
Independent Claim
153 words
A knowledge-based speech recognition system for recognizing an input speech signal comprising vocalic and non-vocalic intervals, each of the vocalic intervals having a pitch period, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal; means for characterizing the series of segments based upon acoustic events detected within the input speech signal to obtain an acoustic feature vector; means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and at least one acoustic event transcription; and means for selecting a word choice by comparing the acoustic feature vector to the acoustic event transcriptions of the multiplicity of words.
Claim 2
Dependent Claim
153 words
A knowledge-based speech recognition system for recognizing an input speech signal comprising vocalic and non-vocalic intervals, each of the vocalic intervals having a pitch period, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal; means for characterizing the series of segments based upon acoustic events detected within the input speech signal to obtain an acoustic feature vector; means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and at least one acoustic event transcription; and means for selecting a word choice by comparing the acoustic feature vector to the acoustic event transcriptions of the multiplicity of words.
Claim 3
Dependent Claim
34 words
The system as defined in claim 1 further comprising means for analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 4
Dependent Claim
34 words
The system as defined in claim 1 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and display means for displaying the word choice.
Claim 5
Dependent Claim
39 words
The system as defined in claim 1 wherein the non-vocalic intervals are computed based upon a frame length corresponding to a mean value of the estimations of the pitch periods for preceding vocalic intervals in the input speech signal.
Claim 6
Dependent Claim
20 words
The system as defined in claim 1 further comprising means for computing formant tracks for the segments including vocalic intervals.
Claim 7
Dependent Claim
53 words
The system as defined in claim 5 wherein the means for computing formant tracks analyzes the series of segments to determine a plurality of maxima within the vocalic intervals of the segments and constructs polyline representations of the formant tracks for each vocalic interval based on spectral amplitude characteristics of the vocalic interval.
Claim 8
Dependent Claim
56 words
The system as defined in claim 5 wherein the means for computing formant tracks analyzes the series of segments to determine a plurality of maxima within the vocalic intervals of the segments and constructs polyline representations of the formant tracks for each vocalic interval based on both spectral amplitude and phase characteristics of the vocalic interval.
Claim 9
Dependent Claim
42 words
The system as defined in claim 1 wherein the means for storing further includes grammatical attributes for at least some of the multiplicity of words, the means for selecting further comprising means for employing the grammatical attributes to determine the word choice.
Claim 10
Dependent Claim
42 words
The system as defined in claim 1 wherein the means for storing further includes semantic attributes for one or more of the multiplicity of words, the means for selecting further comprising means for employing the semantic attributes to determine the word choice.
Claim 11
Dependent Claim
170 words
A knowledge-based speech recognition system for recognizing an input speech signal, the input speech signal including a sequence of phonemes, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; means for characterizing the series of segments based upon acoustic events detected within the input speech signal to obtain an acoustic feature vector; means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and an acoustic event transcription, the acoustic event transcription comprising a lattice of acoustic events corresponding to a plurality of pronunciations; and means for selecting a word choice by comparing the acoustic feature vector to selected lattices of acoustic event transcriptions.
Claim 12
Dependent Claim
170 words
A knowledge-based speech recognition system for recognizing an input speech signal, the input speech signal including a sequence of phonemes, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; means for characterizing the series of segments based upon acoustic events detected within the input speech signal to obtain an acoustic feature vector; means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and an acoustic event transcription, the acoustic event transcription comprising a lattice of acoustic events corresponding to a plurality of pronunciations; and means for selecting a word choice by comparing the acoustic feature vector to selected lattices of acoustic event transcriptions.
Claim 13
Dependent Claim
34 words
The system as defined in claim 10 further comprising means for analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 14
Dependent Claim
34 words
The system as defined in claim 10 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and display means for displaying the word choice.
Claim 15
Dependent Claim
48 words
The system as defined in claim 10 wherein one or more of the multiplicity of words are described by a baseform phonetic transcription, the means for storing the dictionary further comprising means for generating a cloud of phonetic transcriptions by applying phonological rules to the baseform phonetic transcription.
Claim 16
Dependent Claim
36 words
The system as defined in claim 13 wherein, for one or more of the multiplicity of words, each member of a cloud of phonetic transcriptions has an acoustic event transcription including a lattice of acoustic events.
Claim 17
Dependent Claim
46 words
The system as defined in claim 10 wherein one or more of the multiplicity of words are organized into orbits corresponding to a degree of similarity between the acoustic event transcriptions, the means for selecting including means for comparing the acoustic feature vector to selected orbits.
Claim 18
Dependent Claim
58 words
The system as defined in claim 15 wherein the acoustic feature vector includes a plurality of acoustic events and the means for comparing the acoustic feature vector to selected orbits compares individual ones of the acoustic events of the acoustic feature vector to respective ones of the lattice of acoustic events of the selected orbits using three-way logic.
Claim 19
Dependent Claim
35 words
The system as defined in claim 15 wherein the multiplicity of words represents potential candidates for the word choice, the means for comparing reducing the number of potential candidates by about two orders of magnitude.
Claim 20
Dependent Claim
35 words
The system as defined in claim 10 wherein for one or more of the multiplicity of words, the acoustic transcriptions include information relating to the behavior of formants occurring within the one or more words.
Claim 21
Dependent Claim
42 words
The system as defined in claim 10 wherein the means for storing further includes grammatical attributes for one or more of the multiplicity of words, the means for selecting further comprising means for employing the grammatical attributes to determine the word choice.
Claim 22
Dependent Claim
42 words
The system as defined in claim 10 wherein the means for storing further includes semantic attributes for one or more of the multiplicity of words, the means for selecting further comprising means for employing the semantic attributes to determine the word choice.
Claim 23
Dependent Claim
42 words
The system as defined in claim 10 wherein the means for storing further includes syntactic attributes for one or more of the multiplicity of words, the means for selecting further comprising means for employing the syntactic attributes to determine the word choice.
Claim 24
Dependent Claim
41 words
The system as defined in claim 10 wherein phonetic transcription includes a lattice of phonemes and the means for selecting further comprises means for comparing the series of segments of the input speech signal to the lattice of phonemes using dendrograms.
Claim 25
Dependent Claim
82 words
The system as defined in claim 22 wherein the means for storing includes, for each one of the multiplicity of words, a series of values representative of the spectral characteristics and formants behavior for the word, the means for characterizing the series of segments computing a plurality of values representative of the spectral characteristics for the segments, the means for selecting comparing the series of values for selected ones of the multiplicity of words to the plurality of values for the segments.
Claim 26
Dependent Claim
187 words
A knowledge-based speech recognition system for recognizing an input speech signal, the input speech signal including a sequence of phonemes, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; means for characterizing the acoustic and spectral characteristics of the input speech signal: means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a series of acoustic events, spectral characteristics and a series of grammatical, semantic and syntactic attributes; and means for selecting a word choice from amongst the multiplicity of words by weighing, for one or more of the multiplicity of words, correspondence between the acoustic events and the acoustic characteristics, between the spectral characteristics of the words and the input speech signal, and the grammatical, semantic and syntactic attributes of the words.
Claim 27
Dependent Claim
187 words
A knowledge-based speech recognition system for recognizing an input speech signal, the input speech signal including a sequence of phonemes, the system comprising: means for capturing the input speech signal; means for segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; means for characterizing the acoustic and spectral characteristics of the input speech signal: means for storing a dictionary having a multiplicity of words, each one of the multiplicity of words described by a series of acoustic events, spectral characteristics and a series of grammatical, semantic and syntactic attributes; and means for selecting a word choice from amongst the multiplicity of words by weighing, for one or more of the multiplicity of words, correspondence between the acoustic events and the acoustic characteristics, between the spectral characteristics of the words and the input speech signal, and the grammatical, semantic and syntactic attributes of the words.
Claim 28
Dependent Claim
34 words
The system as described in claim 24 further comprising means for analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 29
Dependent Claim
34 words
The system as defined in claim 24 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and display means for displaying the word choice.
Claim 30
Independent Claim
166 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal comprising vocalic and non-vocalic intervals, each of the vocalic intervals having a pitch period, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal; generating an acoustic feature vector by characterizing the series of segments based upon acoustic events detected within the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and at least one acoustic event transcription; and comparing the acoustic feature vector to the acoustic event transcriptions of the multiplicity of words to select a word choice.
Claim 31
Independent Claim
166 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal comprising vocalic and non-vocalic intervals, each of the vocalic intervals having a pitch period, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal; generating an acoustic feature vector by characterizing the series of segments based upon acoustic events detected within the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and at least one acoustic event transcription; and comparing the acoustic feature vector to the acoustic event transcriptions of the multiplicity of words to select a word choice.
Claim 32
Dependent Claim
35 words
The method as defined in claim 27 further comprising a step of analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 33
Dependent Claim
38 words
The method as defined in claim 27 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and the method further comprises a step of displaying the word choice.
Claim 34
Dependent Claim
41 words
The method as defined in claim 27 wherein the step of segmenting further comprises computing a frame length for non-vocalic intervals as a mean value of the estimations of the pitch periods for preceding vocalic intervals in the input speech signal.
Claim 35
Dependent Claim
21 words
The method as defined in claim 27 further comprising the step of computing formant tracks for the segments including vocalic intervals.
Claim 36
Dependent Claim
54 words
The method as defined in claim 31 wherein the step of computing formant tracks comprises analyzing the series of segments to determine a plurality of maxima within the vocalic intervals of the segments and constructing polyline representations of the formant tracks for each vocalic interval based on spectral amplitude characteristics of the vocalic interval.
Claim 37
Dependent Claim
56 words
The method as defined in claim 31 wherein the means for computing formant tracks analyzes the series of segments to determine a plurality of maxima within the vocalic intervals of the segments and constructs polyline representations of the formant tracks for each vocalic interval based on both spectral amplitude and phase characteristics of the vocalic interval.
Claim 38
Dependent Claim
38 words
The method as defined in claim 27 wherein the dictionary further includes grammatical attributes for one or more of the multiplicity of words, the step of comparing further comprising employing the grammatical attributes to determine the word choice.
Claim 39
Dependent Claim
38 words
The method as defined in claim 27 wherein the dictionary further includes semantic attributes for one or more of the multiplicity of words, the step of comparing further comprising employing the semantic attributes to determine the word choice.
Claim 40
Independent Claim
179 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal including a sequence of phonemes, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; generating an acoustic feature vector by characterizing the series of segments based upon acoustic events detected within the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and an acoustic event transcription, the acoustic event transcription comprising a lattice of acoustic events corresponding to a plurality of pronunciations; and comparing the acoustic feature vector to selected lattices of acoustic event transcriptions to select a word choice.
Claim 41
Independent Claim
179 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal including a sequence of phonemes, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; generating an acoustic feature vector by characterizing the series of segments based upon acoustic events detected within the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a phonetic transcription and an acoustic event transcription, the acoustic event transcription comprising a lattice of acoustic events corresponding to a plurality of pronunciations; and comparing the acoustic feature vector to selected lattices of acoustic event transcriptions to select a word choice.
Claim 42
Dependent Claim
35 words
The method as defined in claim 36 further comprising a step of analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 43
Dependent Claim
38 words
The method as defined in claim 36 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and the method further comprises a step of displaying the word choice.
Claim 44
Dependent Claim
45 words
The method as defined in claim 36 wherein one or more of the multiplicity of words are described by a baseform phonetic transcription, the method further comprising a step of generating a cloud of phonetic transcriptions by applying phonological rules to the baseform phonetic transcription.
Claim 45
Dependent Claim
36 words
The method as defined in claim 39 wherein, for one or more of the multiplicity of words, each member of a cloud of phonetic transcriptions has an acoustic event transcription including a lattice of acoustic events.
Claim 46
Dependent Claim
44 words
The method as described in claim 36 wherein one or more of the multiplicity of words are organized into orbits corresponding to a degree of similarity between the acoustic event transcriptions, the step of comparing comprising comparing the acoustic feature vector to selected orbits.
Claim 47
Dependent Claim
58 words
The method as defined in claim 41 wherein the acoustic feature vector includes a plurality of acoustic events and the step of comparing the acoustic feature vector to selected orbits compares individual ones of the acoustic events of the acoustic feature vector to respective ones of the lattice of acoustic events of the selected orbits using three-way logic.
Claim 48
Dependent Claim
35 words
The method as defined in claim 41 wherein the multiplicity of words represents potential candidates for the word choice, the step of comparing reducing the number of potential candidates by about two orders of magnitude.
Claim 49
Dependent Claim
35 words
The method as defined in claim 36 wherein for one or more of the multiplicity of words, the acoustic transcriptions include information relating to the behavior of formants occurring within the one or more words.
Claim 50
Dependent Claim
37 words
The method as defined in claim 36 wherein the dictionary further includes grammatical attributes one or more of the multiplicity of words, the step of comparing further comprising employing the grammatical attributes to determine the word choice.
Claim 51
Dependent Claim
38 words
The method as defined in claim 36 wherein the dictionary further includes semantic attributes for one or more of the multiplicity of words, the step of comparing further comprising employing the semantic attributes to determine the word choice.
Claim 52
Dependent Claim
38 words
The method as defined in claim 36 wherein the dictionary further includes syntactic attributes for one or more of the multiplicity of words, the step of comparing further comprising employing the syntactic attributes to determine the word choice.
Claim 53
Dependent Claim
39 words
The method as defined in claim 36 wherein phonetic transcription includes a lattice of phonemes and the step of comparing further comprises comparing the series of segments of the input speech signal to the lattice of phonemes using dendrograms.
Claim 54
Dependent Claim
88 words
The method as defined in claim 48 wherein the dictionary includes, for each one of the multiplicity of words, a series of values representative of the spectral characteristics and formants behavior for the word, the step of segmenting the input speech signal into a series of segments comprises computing a plurality of values representative of the spectral characteristics for the segments, and the step of comparing comprises comparing the series of values for selected ones of the multiplicity of words to the plurality of values for the segments.
Claim 55
Independent Claim
196 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal including a sequence of phonemes, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; computing acoustic and spectral characteristics representative of the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a series of acoustic events, spectral characteristics and a series of grammatical, semantic and syntactic attributes; and selecting a word choice from amongst the multiplicity of words by weighing, for one or more of the multiplicity of words, correspondence between the acoustic events and the acoustic characteristics, between the spectral characteristics of the words and the input speech signal, and the grammatical, semantic and syntactic attributes of the words.
Claim 56
Independent Claim
196 words
A method of applying knowledge-based rules in a processor-based system to recognize an input speech signal as a word in a natural language, the input speech signal including a sequence of phonemes, the method comprising steps of: capturing and storing the input speech signal; segmenting the input speech signal into a series of segments including vocalic intervals and non-vocalic intervals, the vocalic intervals having a frame length computed based on an estimation of the pitch period of the vocalic intervals of the input speech signal, the series of segments approximately corresponding to the sequence of phonemes; computing acoustic and spectral characteristics representative of the input speech signal; retrieving from a store a dictionary having a multiplicity of words, each one of the multiplicity of words described by a series of acoustic events, spectral characteristics and a series of grammatical, semantic and syntactic attributes; and selecting a word choice from amongst the multiplicity of words by weighing, for one or more of the multiplicity of words, correspondence between the acoustic events and the acoustic characteristics, between the spectral characteristics of the words and the input speech signal, and the grammatical, semantic and syntactic attributes of the words.
Claim 57
Dependent Claim
35 words
The method as defined in claim 50 further comprising a step of analyzing the input speech signal to determine a sequence of word choices based upon syntactic and semantic attributes of the selected word choices.
Claim 58
Dependent Claim
38 words
The method as defined in claim 50 wherein the dictionary further comprises at least one orthographic word corresponding to each one of the multiplicity of words and the method further comprises a step of displaying the word choice.
References Cited (Backward Citations)
US3700815A
Cited
US4718094A
Cited
US4748670A
Cited
US4903305A
Cited
US5040127A
Cited
US4837831A
Cited
US5027406A
Cited
US5033087A
Cited
US5170432A
Cited
US5202952A
Cited
US5526463A
Cited
US5384892A
Cited
US5390279A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2001-08-24 | 2002-02-23 | 2002-08-25 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2005-08-24 | 2006-02-23 | 2006-08-25 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2009-08-24 | 2010-02-23 | 2010-08-25 | $7700.00 | Window Closed / Lapsed |