EP0216118A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Noise compensation in speech recognition apparatus
Assignee: International Standard Electric Corp
•
Filing: 1986-08-15
•
Granted: —
•
Expires: 2006-08-15
Abstract
Abstract A method and apparatus for noise suppression for speech recognition systems which employs the principle of a least mean square estimation which is implemented with conditional expected values. Essentially, according to this method, one computes a series of optimal estimators which estimators and their variances are then employed to implement a noise immune metric. This noise immune metric enables the system to substitute a noisy distance with an expected value which value is calculated according to combined speech and noise data which occurs in the bandpass filter domain. Thus the system can be used with any set of speech parameters and is relatively independent of a specific speech recognition apparatus structure.
Inventors
- 👤 Steven Frank Boll
- 👤 Jack Elliott Porter
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-10-26 20:32
CPC Classifications
G; G10; G10L; G10L15/00; G10L15/20
Claims Analysis
Total Claims: 13
Claim 1
Independent Claim
165 words
A method of reducing noise in a speech recognition apparatus by using the minimum mean square error estimate of the actual squared Euclidean distance between template speech and unknown speech signals as conditioned by noisy observations, comprising the steps of: providing a first value indicative of the value of noise-free template, t,, providing a second value indicative of the value of noise free unknown speech signal u s , providing a third value indicative of a noisy template t s+n , providing a fourth value indicative of a noisy unknown speech signal u s+n , providing a fifth value indicative of the average power of said unknown speech signals P s , providing a sixth value indicative of the average power of noise signals P n , computing a new expected distance between said template speech to said unknown speech signals according to the following algorithm: using said new expected distance as computed to measure the similarity between said template speech and said unknown speech.
Claim 2
Independent Claim
165 words
A method of reducing noise in a speech recognition apparatus by using the minimum mean square error estimate of the actual squared Euclidean distance between template speech and unknown speech signals as conditioned by noisy observations, comprising the steps of: providing a first value indicative of the value of noise-free template, t,, providing a second value indicative of the value of noise free unknown speech signal u s , providing a third value indicative of a noisy template t s+n , providing a fourth value indicative of a noisy unknown speech signal u s+n , providing a fifth value indicative of the average power of said unknown speech signals P s , providing a sixth value indicative of the average power of noise signals P n , computing a new expected distance between said template speech to said unknown speech signals according to the following algorithm: using said new expected distance as computed to measure the similarity between said template speech and said unknown speech.
Claim 3
Dependent Claim
18 words
The method according to claim 1, wherein said values are provided at specific frequencies within the speech band.
Claim 4
Dependent Claim
19 words
The method according to claim 2, wherein said frequencies employed are at 300, 425, 1063, 2129 and 3230 Hz.
Claim 5
Dependent Claim
16 words
The method according to claim 3, wherein said values are provided at selected average signal-to-noise ratios.
Claim 6
Dependent Claim
19 words
The method according to claim 4, wherein said average signal-to-noise ratios are 0 db, 10 db, and 20 db.
Claim 7
Dependent Claim
47 words
The method according to claim 1, wherein said values are replaced by mean values to provide a new expected distance equal to: where t + u. are the expected values of the template and unknown and a j , a u are the variances of the estimates.
Claim 8
Dependent Claim
19 words
The method according to claim 1, including the step storing a plurality of said first values in a memory.
Claim 9
Dependent Claim
29 words
The method according to claim 1, including the steps storing a plurality of any one or more of said second, third, fourth, fifth and sixth values in a memory.
Claim 10
Dependent Claim
23 words
The method according to claim 7, wherein said values stored are indicative of said first value at different frequencies within said speech bandwidth.
Claim 11
Dependent Claim
20 words
The method according to claim 7, wherein said values stored are indicative of said first value at different signal-to-noise ratios.
Claim 12
Independent Claim
135 words
A method of compensating for noisy signals for use by a speech recognition apparatus of the type which measures the similarity between segments of an unknown speech signal and template speech by computing the Euclidean distance between respective segment parameters of the unknown speech and the template speech comprising the steps of: computing the optimal least squares estimator at each frequency contained in said unknown and template speech signals, providing a variance for each optimal estimator computed, replacing the speech segments and the template segments with said computed optimal estimators, providing a new squared Euclidean distance from said replaced estimators, adding to said new Euclidean distance said variances for each optimal estimator to provide a new distance indicative of an expected distance value, and using said new distance to measure the similarity between said segments.
Claim 13
Independent Claim
135 words
A method of compensating for noisy signals for use by a speech recognition apparatus of the type which measures the similarity between segments of an unknown speech signal and template speech by computing the Euclidean distance between respective segment parameters of the unknown speech and the template speech comprising the steps of: computing the optimal least squares estimator at each frequency contained in said unknown and template speech signals, providing a variance for each optimal estimator computed, replacing the speech segments and the template segments with said computed optimal estimators, providing a new squared Euclidean distance from said replaced estimators, adding to said new Euclidean distance said variances for each optimal estimator to provide a new distance indicative of an expected distance value, and using said new distance to measure the similarity between said segments.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.