FireRedTtsStage
Enums · AiDotNet.TextToSpeech.CodecBased
Which of FireRedTTS’s networks a training step updates.
Fields
| Field | Summary |
|---|---|
FireRedTtsStage AcousticLanguageModel |
The streamable decoder’s multi-stream acoustic language model (§3.3.2), on a recording’s semantic tokens and its Mel codec streams. |
FireRedTtsStage FlowDecoder |
The flow-matching Mel decoder (§3.3.1), on a recording’s tokens and 16 kHz mel, with the timbre from another recording of the speaker. |
FireRedTtsStage MelCodec |
The streamable decoder’s Mel codec (§3.3.2), against its mel discriminator in alternate turns. |
FireRedTtsStage SpeechTokenizer |
The semantic-aware speech tokenizer (§3.1, eq. |
FireRedTtsStage TextToSpeech |
The text-to-speech language model (§3.2), on a recording’s semantic tokens and speaker embedding. |
int value__ |