Emilia-YODAS-Voice-Conversion

We sample https://huggingface.co/datasets/amphion/Emilia-Dataset YODAS set for voice conversion.

1. Filter transcriptions based on character repetitiveness and word ngrams.
2. Filter speaker similarity using https://huggingface.co/nvidia/speakerverification_en_titanet_large during speaker permutation.
3. Convert audio to speech tokens using https://huggingface.co/neuphonic/neucodec

We also upload the full permutation as zip files.

Speech Tokenizer

Convert audio to speech tokens using https://huggingface.co/neuphonic/neucodec 50Hz, with total 5.7B speech tokens.

Statistics

1. DE, 5558.53 hours.
2. EN, 13493.57 hours.
3. FR, 6954.43 hours.
4. JA, 1120.36 hours.
5. KO, 6991.33 hours.
6. ZH, 326.01 hours.