FishSpeechRecipe

Enums · AiDotNet.TextToSpeech.CodecBased

Which training recipe a Fish Speech model’s optimizer follows.

Fields

Field Summary
FishSpeechRecipe Paper The paper’s pre-training (Liao et al.
FishSpeechRecipe ReleaseFineTune The release’s fine-tuning (text2semantic_finetune.yaml): AdamW (0.9, 0.95, 1e-5) at 1e-4, no weight decay, constant after 10 warm-up steps.
int value__