Higgs TTS 3
此 Soniqo 页面记录本地 speech-swift / speech-core 实现中的 Higgs TTS 3。Hugging Face 包链接放在集成说明之后。
先进入站内页面
首页卡片和文档菜单先指向这里；源模型和权重包链接仍在本页提供。
概览
| 模型 | Higgs TTS 3 |
|---|---|
| 用途 | Conversational TTS and zero-shot cloning with inline emotion, style, and SFX tags |
| 后端 | MLX bf16 backbone with float32 codec |
| 输出 | 24 kHz mono waveform, 25 frames per second |
| 语言 | 100+ languages |
| 许可证 | Boson Higgs TTS 3 Research and Non-Commercial License |
| 状态 | Ready through speech speak --engine higgs and the HiggsTTS Swift product |
| 来源 | Boson AI higgs-tts-3-4b |
| Swift 产品 | HiggsTTS |
| CLI / 运行时 | speech speak --engine higgs |
使用
下面的片段对应当前 speech-swift 仓库暴露的 API 或命令。
# Expressive cloned speech with an inline emotion tag.
.build/release/speech speak "<|emotion:elation|> We did it — the demo works!" \
--engine higgs \
--voice-sample reference.wav \
--higgs-ref-text "Text spoken in the reference clip." \
-o higgs.wav模型链接
实现说明
- RTF 0.78 on an M5 Pro for cloned and plain synthesis alike: sampling stays on the GPU with the delay ramp masked on-device, and asyncEval overlaps decode with the end-of-codes state machine.
- Cloning gates: English lexical overlap 1.00 and Mandarin CER 0.000 on ASR roundtrips; the language model and codec are parity-validated against the upstream implementations in both directions.
- Control tags are ordinary tokens inside the target text: <|emotion:...|>, <|style:whispering|>, <|sfx:laughter|>, <|prosody:pause|> and friends.
- Encoding a 12 s reference costs about 0.6 s once and the codes are reusable across generations; omit --voice-sample for reference-free synthesis in a stock voice.
- The CLI defaults to temperature 0.8 because the upstream 1.0 wobbles on single sentences; expect about 8.6 GB peak RSS for the 4B backbone.