Detailed calculation methods for pitch and speaking speed

line / LibriTTS-P

LibriTTS-P: A Corpus with Speaking Style and Speaker Identity Prompts for Text-to-Speech and Style Captioning

112 stars 2 forks source link

Detailed calculation methods for pitch and speaking speed #3

Open hoang1007 opened 1 month ago

hoang1007 commented 1 month ago

Hello @MasayaKawamura

Thank you for the LibriTTS-P dataset release and paper. I am using your dataset to training my model. Can you please provide detailed method to calculate pitch and speaking speed for each utterance in the dataset?

Thanks!

r9y9 commented 1 week ago

Hi, sorry for being late. We've open-sourced our code for PromptTTS++ with the full details of style prompt generation. Please check https://github.com/line/promptttspp/tree/main/data_prep for details. For instance, https://github.com/line/promptttspp/blob/3e6bd0eaa7d0bfadb5f33a530726dd78efc748dd/data_prep/compute_utt_stats.py#L32-L61 is the code to compute speaking speed.