Normalization of speech by adaptive labelling

• US 4,926,488 A
• Filed: 07/09/1987
• Issued: 05/15/1990
• Est. Priority Date: 07/09/1987
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Nahamoo, David, Nadas, Arthur J.
• Primary Examiner(s): Harkcom, Gary V.
• Assistant Examiner(s): Knepper, David D.
• Application Number:
  US07/071,687
• Time in Patent Office: 1,041 Days
• Field of Search: 364/513.5, 381/41-50
• US Class Current: 704/233
• CPC Class Codes: G10L 15/07 to the speaker
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence

Abstract

In a speech processor system in which prototype vectors of speech are generated by an acoustic processor under reference noise and known ambient conditions and in which feature vectors of speech are generated during varying noise and other ambient and recording conditions, normalized vectors are generated to reflect the form the feature vectors would have if generated under the reference conditions. The normalized vectors are generated by: (a) applying an operator function A_i to a set of feature vectors x occurring at or before time interval i to yield a normalized vector y_i =A_i (x); (b) determining a distance error vector E_i by which the normalized vector is projectively moved toward the closest prototype vector to the normalized vector y_i ; (c) up-dating the operator function for next time interval to correspond to the most recently determined distance error vector; and (d) incrementing i to the next time interval and repeating steps (a) through (d) wherein the feature vector corresponding to the incremented i value has the most recent up-dated operator function applied thereto. With successive time intervals, successive normalized vectors are generated based on a successively up-dated operator function. For each normalized vector, the closest prototype thereto is associated therewith. The string of normalized vectors or the string of associated prototypes (or respective label identifiers thereof) or both provide output from the acoustic processor.

Claims

1. A speech coding apparatus comprising:

means for measuring the value of at least one feature of an utternace, said utternace occurring over a series of successive time intervals, said means measuring the feature value of the utterance during each time interval to produce a series of feature vector signals representing the feature values;

means for storing a plurality of prototype vector signals, each prototype vector signal having at least one parameter value and having a unique identification value;

means for generating a first modified feature vector signal having a modified feature value, said modified feature value being related, by a modification function, to the feature value of a first feature vector signal in the series of feature vector signals;

means for comparing the modified feature value of the first modified feature vector signal to the parameter values of the prototype vector signals to determine the associated prototype vector signal which is best matched to the first modified feature vector signal;

means for altering the modification function to improve the match between the modified feature vector signal and its associated prototype vector signal determined by the comparison;

means for generating a second modified feature vector signal having a modified feature value, said modified feature value of the second modified feature vector being related, by the altered modification function, to the feature value of a second feature vector signal in the series of feature vector signals, said second feature vector signal following the first feature vector signal;

means for comparing the modified feature value of the second modified feature vector signal to the parameter values of the prototype vector signals to determine the associated prototype vector signal which is best matched to the second modified feature vector signal; and

means for outputting the identification value of the prototype vector signal associated with the second modified feature vector as a coded representation of the second feature vector signal.

2. An apparatus as claimed in claim 1, characterized in that the second feature vector signal immediately follows the first feature vector signal.

3. An apparatus as claimed in claim 2, characterized in that the modification function and the altered modification function normalize the feature vector signals.

4. An apparatus as claimed in claim 3, characterized in that each means for comparing determines the prototype vector signal which is closest to the modified feature vector signal.

5. A method of coding speech, said method comprising the steps of:

measuring the value of at least one feature of an utterance, said utternace occurring over a series of successive time intervals, the feature value of the utterance being measured during each time interval to produce a series of feature vector signals representing the feature values;

storing a plurality of prototype vector signals, each prototype vector signal having at least one parameter value and having a unique identification value;

generating a first modified feature vector signal having a modified feature value, said modified feature value being related, by a modification function, to the feature value of a first feature vector signal in the series of feature vector signals;

comparing the modified feature value of the first modified feature vector signal to the parameter values of the prototype vector signals to determine the associated prototype vector signal which is best matched to the first modified feature vector signal;

altering the modification function to improve the match between the modified feature vector signal and its associated prototype vector signal determined by the comparison;

generating a second modified feature vector signal having a modified feature value, said modified feature value of the second modified feature vector being related, by the altered modification function, to the feature value of a second feature vector signal in the series of feature vector signals, said second feature vector signal following the first feature vector signal;

comparing the modified feature value of the second modified feature vector signal to the parameter values of the prototype vector signals to determine the associated prototype vector signal which is best matched to the second modified feature vector signal; and

outputting the identification value of the prototype vector signal associated with the second modified feature vector as a coded representation of the second feature vector signal.

6. A method as claimed in claim 5, characterized in that the second feature vector signal immediately follows the first feature vector signal.

7. A method as claimed in claim 6, characterized in that the modification function and the altered modification function normalize the feature vector signals.

8. An method as claimed in claim 7, characterized in that each step of comparing determines the prototype vector signal which is closest to the modified feature vector signal.