YouVersion African Speech — MOSS-TTS-Nano prepped
Domaine:
natural language processingType de record:
dataset
Créateur:
Afr
Hôte:
Preprocessed training data for finetuning MOSS-TTS-Nano:
every audio clip from AfriSpeech/youversion-african-speech
has been encoded into discrete audio codes with
MOSS-Audio-Tokenizer-Nano, so you can train
directly with the MOSS-TTS-Nano finetuning pipeline
without downloading or re-encoding the ~63 GB of source audio.
What's in it