ComfyUI Node
Spark_TTS_Clone
A ComfyUI node in ComfyUI_Spark_TTS with 12 inputs and 2 outputs.
Spark_TTS_Clone
- Audio_reference
- Audio
- Node Status
◄textCloning a voice with Spark TTS is interesting.►
◄custom_prompt_text►
◄speaker_presetad_male_en►
◄pitchmoderate►
◄speedmoderate►
◄temperature0.70►
◄top_k30►
◄top_p0.70►
◄max_new_tokens2020►
◄keep_model_loadedtrue►
◄use_cpufalse►
CategoryComfyUI_Spark_TTS
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| text | STRING | Cloning a voice with Spark TTS is interesting. | Text to be synthesized / 待合成的文本 |
| custom_prompt_text | STRING | Optional text transcription for the custom audio reference, improves cloning accuracy / 自定义参考音频的文本转录（可选），提高克隆准确性 | |
| speaker_preset | COMBO | ad_male_en | Select a preset speaker for cloning / 选择一个预设说话人进行克隆 |
| pitch | COMBO | moderate | Output voice pitch level (e.g., moderate). Effect depends on cloning strength / 输出语音的音高水平（例如：中等）。效果取决于克隆强度 |
| speed | COMBO | moderate | Output voice speed level (e.g., moderate). Effect depends on cloning strength / 输出语音的语速水平（例如：中等）。效果取决于克隆强度 |
| temperature | FLOAT | 0.700–1 | Sampling temperature for generation / 生成的采样温度 |
| top_k | INT | 300–100 | Top-K sampling parameter / Top-K 采样参数 |
| top_p | FLOAT | 0.700–1 | Top-P (nucleus) sampling parameter / Top-P (核) 采样参数 |
| max_new_tokens | INT | 2020100–90000 | Maximum number of new tokens to generate / 要生成的最大新 token 数 |
| keep_model_loaded | BOOLEAN | true | Keep model in VRAM after use for faster subsequent runs / 使用后将模型保留在显存中以便后续运行更快 |
| use_cpu | BOOLEAN | false | Force node execution on CPU instead of GPU / 强制节点在 CPU 而非 GPU 上执行 |
| Audio_referenceopt | AUDIO | Custom audio file for voice cloning. Overrides 'speaker_preset' / 用于语音克隆的自定义音频文件。会覆盖 'speaker_preset' |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| Audio | AUDIO | — |
| Node Status | STRING | — |