System and method for outlier identification to remove poor alignments in speech synthesis

1. A method for identifying outlying results in audio files used for model training Hidden Markov Models, in a text-to-speech system, applying fundamental frequency, the method comprising the steps of:

a. extracting a plurality of values from the audio files of a text-to-speech system, wherein the values comprise fundamental frequencies;

b. generating alignments using the extracted plurality of values from the audio files in the text-to-speech system;

c. separating out instances of phonemes;

d. determining, for each separated instance, an average fundamental frequency value and an average duration value;

e. identifying an instance as an outlier, wherein an outlier is identified if:

i. the phoneme is a vowel;

ii. the average fundamental frequency of an instance is less than a predetermined value;

iii. the duration of the instance is greater than twice the average duration of a phoneme; and

iv. the duration of the instance is less than half of the average duration of a phoneme; and

f. determining a total number of the instances identified as outliers from step (e) for each sentence in the audio files;

g. discarding a sentence in the audio files in the text-to-speech system from training a Hidden Markov Model if the total number of the instances identified as outliers for the sentence is greater than a predetermined sentence outlier threshold.

2. The method of claim 1 , wherein the extracting is performed using a pitch tracking tool.

3. The method of claim 1 , wherein the generating is performed using a speech recognition system.

4. The method of claim 1 , wherein the alignments are generated at a phoneme level.

5. The method of claim 1 , wherein the threshold of outliers is

a predetermined value, and

the predetermined value of the threshold is empirically chosen.

6. The method of claim 1 , wherein the number of outliers of step (f) is empirically chosen.

7. The method of claim 6 , wherein the number of outliers is five.

8. A method for identifying outlying results in audio files used for model training Hidden Markov Models, in a text-to-speech system, applying group delay algorithms, the method comprising the steps of:

a. generating alignments of the audio files of the text-to-speech system at a phoneme level;

b. generating alignments of the audio files of the text-to-speech system at a syllable level;

c. adjusting the alignments at the syllable level using the group delay algorithms;

d. separating each syllable from the audio files of the text-to-speech system into a separate audio file;

e. generating, for each separate audio file of the text-to-speech system, phonemes of the separate audio files using phoneme boundaries for each syllable and an existing phoneme model;

f. determining a likelihood value of each generated phoneme, wherein if the likelihood value meets a criteria, identifying the generated phoneme as an outlier; and

g. determining a total number of outliers for each sentence in the audio files in the text-to-speech system;

h. discarding a sentence in the audio files in the text-to-speech system from training a Hidden Markov Model if the total number of the instances identified as outliers for the sentence is greater than a predetermined sentence outlier threshold.

9. The method of claim 8 , wherein the generating of step (a) is performed using a speech recognition system.

10. The method of claim 8 , wherein the generating of step (b) is performed using at least one of: a speech recognition system and a phoneme model.

11. The method of claim 8 , wherein the criteria comprises a small value.

12. The method of claim 8 , wherein the criteria comprises a failed alignment.

13. The method of claim 8 , wherein the number of outliers of step (g) comprises three.

14. A method for synthesizing speech in a text-to-speech system, wherein the system comprises at least a speech database, a database capable of storing Hidden Markov Models, and a synthesis filter, the method comprising the steps of:

a. identifying outlying results in audio files from the speech database and removing the outlying results before model training, comprising the steps of:

i. extracting a plurality of values from the audio files, wherein the values comprise fundamental frequencies,

ii. generating alignments using the extracted plurality of values from the audio files,

iii. separating out instances of phonemes,

iv. determining, for each separated instance, an average fundamental frequency value and an average duration value,

v. identifying an instance as an outlier, wherein an outlier is identified if:

(a) the phoneme is a vowel;

(b) the average fundamental frequency of an instance is less than a predetermined value;

(c) the duration of the instance is greater than twice the average duration of a phoneme; and

(d) the duration of the instance is less than half of the average duration of a phoneme, and

vi. determining a total number of the instances identified in step (a)(v) as outliers for each sentence in the audio files;

vii. discarding a sentence in the audio files from model training if the total number of the instances identified as outliers for the sentence is greater than a predetermined sentence outlier threshold,

b. converting a speech signal from the speech database into parameters and extracting the parameters from the speech signal;

c. training Hidden Markov Models using the extracted parameters from the speech signal and using the labels from the speech database to produce context dependent Hidden Markov Models;

d. storing the context dependent Hidden Markov Models in the database capable of storing Hidden Markov Models;

e. inputting text and analyzing the text, wherein said analyzing comprises extracting labels from the text;

f. utilizing said labels to generate parameters from the context dependent Hidden Markov Models;

g. generating an other signal from the parameters;

h. inputting the other signal and the parameters into the synthesis filter; and

i. producing synthesized speech as the other signal passes through the synthesis filter.

15. The method of claim 14 , wherein the parameters of step (b) comprise one or more of excitation and spectral.

16. The method of claim 14 , wherein the parameters of step (f) comprise one or more of excitation and spectral.

17. The method of claim 14 , wherein the extracting is performed using a pitch tracking tool.

18. The method of claim 14 , wherein the generating is performed using a speech recognition system.