Systems and methods for implementing segmentation in speech recognition systems
View Patent ↗A speech recognition system ( 105 ) includes an acoustic front end ( 115 ) and a processing unit ( 125 ). The acoustic front end ( 115 ) receives frames of acoustic data and determines cepstral coefficients for each of the received frames. The processing unit ( 125 ) determines a number of peaks in the cepstral coefficients for each of the received frames of acoustic data and compares the peaks in the cepstral coefficients of a first one of the received frames with the peaks in the cepstral coefficients of at least a second one of the received frames. The processing unit ( 125 ) then segments the received frames of acoustic data based on the comparison.
1. A method of recognizing patterns in acoustic data, comprising:
receiving frames of acoustic data;
determining first cepstral coefficients corresponding to the received frames of acoustic data;
determining second cepstral coefficients corresponding to the received frames of acoustic data;
determining a first number of peaks in the first cepstral coefficients and a second number of peaks in the second cepstral coefficients;
determining at least one weighting parameter based on the determined second number of peaks; and
recognizing patterns in the received frames of acoustic data using the at least one weighting parameter.
2. The method of claim 1 , further comprising:
comparing the determined first and second number of peaks.
3. The method of claim 2 , wherein the recognizing patterns in the frames of acoustic data is based on the comparison.
4. The method of claim 1 , further comprising:
determining, based on the received frames of acoustic data, recognition hypothesis scores using a Hidden Markov Model.
5. The method of claim 4 , further comprising:
modifying the recognition hypothesis scores based on the at least one weighting parameter.
6. The method of claim 5 , wherein the recognizing of the patterns in the frames of acoustic data further uses the modified recognition hypothesis scores.
7. The method of claim 5 , further comprising:
re-ordering the modified recognition hypothesis scores.
8. The method of claim 7 , wherein the recognizing of the patterns in the frames of acoustic data further uses the re-ordered modified recognition hypothesis scores.
9. A speech recognition system, comprising:
means for receiving frames of acoustic data;
means for determining a first number of peaks of cepstral coefficients corresponding to the received frames of acoustic data;
means for determining a second number of peaks of cepstral coefficients corresponding to the received frames of acoustic data;
means for determining at least one weighting parameter based on the determined second number of peaks; and
means for recognizing patterns in the frames of acoustic data using the at least one weighting parameter.