SeedTTSTrainingComponent
Enums · AiDotNet.TextToSpeech.CodecBased
The independently trainable part of the Seed-TTS pipeline.
Fields
| Field | Summary |
|---|---|
SeedTTSTrainingComponent AcousticVocoder |
The separately trained acoustic vocoder. |
SeedTTSTrainingComponent SpeechTokenizer |
End-to-end adversarial speech-tokenizer pretraining before token-LM training. |
SeedTTSTrainingComponent TimbreSelfDistillation |
The paper’s timbre-factorization self-distillation objective. |
SeedTTSTrainingComponent TokenDiffusion |
The coarse-to-fine diffusion transformer that turns speech tokens into continuous vocoder features. |
SeedTTSTrainingComponent TokenLanguageModel |
The causal transformer that predicts speech tokens from text and an optional speech prompt. |
int value__ |