I want to use this repo to clone my voice. How do I create the perfect training data set? Here are some ideas I have:
-Read a set of words that contains every letter in the alphabet
-Read a set of words in a normal tone, then read those same words with more energy/power, then read the same with more inflection, then sarcasm, then with inquisitiveness, etc.
-Read continuously with no long pauses
-Read words for a minimum of 3 minutes or X number of words, where X is some relatively large number
-Minimize external noise, so maybe reserve some time in a professional audio production studio
@CorentinJ any tips here on how to make the best possible training data set?
I want to use this repo to clone my voice. How do I create the perfect training data set? Here are some ideas I have:
-Read a set of words that contains every letter in the alphabet -Read a set of words in a normal tone, then read those same words with more energy/power, then read the same with more inflection, then sarcasm, then with inquisitiveness, etc. -Read continuously with no long pauses -Read words for a minimum of 3 minutes or X number of words, where X is some relatively large number -Minimize external noise, so maybe reserve some time in a professional audio production studio
@CorentinJ any tips here on how to make the best possible training data set?