FishAudioReference
An inline reference audio sample for instant voice cloning.
Defined in: src/providers/tts/fishaudio/FishAudioTTS.ts:79
An inline reference audio sample for instant voice cloning.
Remarks
Reference audio is sent as raw binary inside the msgpack request body — this is why FishAudioTTS requires @msgpack/msgpack instead of JSON (which would need base64 inflation). Best results come from a clean 10-30s sample; no training step is required.
Properties
| Property | Type | Description | Defined in |
|---|---|---|---|
audio | ArrayBuffer | Uint8Array<ArrayBufferLike> | Raw audio bytes of the reference sample (e.g., a WAV file’s contents). | src/providers/tts/fishaudio/FishAudioTTS.ts:81 |
text | string | Transcript of the reference audio sample. | src/providers/tts/fishaudio/FishAudioTTS.ts:84 |