ComfyUI Node
IndexTTS-2 TTS
A ComfyUI node in AIIA/Synthesis with 20 inputs and 1 output.
IndexTTS-2 TTS
- indextts_model
- reference_audio
- emotion_audio
- audio
◄text你好，这是一段 IndexTTS-2 语音合成测试。►
◄voice_presetFemale_HQ►
◄emo_alpha1.00►
◄happy0.00►
◄angry0.00►
◄sad0.00►
◄afraid0.00►
◄disgusted0.00►
◄melancholic0.00►
◄surprised0.00►
◄calm0.00►
◄use_emo_textfalse►
◄emo_text►
◄interval_silence200►
◄max_text_tokens_per_segment120►
◄use_randomfalse►
◄seed0►
CategoryAIIA/Synthesis
Inputs (20)
| Name | Type | Default | Description |
|---|---|---|---|
| indextts_model | INDEXTTS_MODEL | — | |
| text | STRING | 你好，这是一段 IndexTTS-2 语音合成测试。 | — |
| voice_preset | COMBO | Female_HQ | 4 options: Female_HQ, Male_HQ, Female, Male |
| reference_audioopt | AUDIO | Speaker voice reference. Leave empty to use voice_preset. | |
| emotion_audioopt | AUDIO | Optional emotion reference audio (separate from speaker voice). | |
| emo_alphaopt | FLOAT | 1.000–1 | Emotion blending strength (0=no emotion, 1=full emotion). |
| happyopt | FLOAT | 0.000–1 | — |
| angryopt | FLOAT | 0.000–1 | — |
| sadopt | FLOAT | 0.000–1 | — |
| afraidopt | FLOAT | 0.000–1 | — |
| disgustedopt | FLOAT | 0.000–1 | — |
| melancholicopt | FLOAT | 0.000–1 | — |
| surprisedopt | FLOAT | 0.000–1 | — |
| calmopt | FLOAT | 0.000–1 | — |
| use_emo_textopt | BOOLEAN | false | Auto-detect emotion from text using built-in Qwen emotion model. Overrides emotion sliders. |
| emo_textopt | STRING | Custom emotion text prompt (used with use_emo_text). Leave empty to use main text. | |
| interval_silenceopt | INT | 2000–2000 | Silence duration (ms) inserted between text segments for long text. |
| max_text_tokens_per_segmentopt | INT | 12030–500 | Max tokens per text segment. Lower = more segments, higher = longer per-segment generation. |
| use_randomopt | BOOLEAN | false | Enable random sampling (reduces voice cloning fidelity). |
| seedopt | INT | 0-1–2147483647 | Random seed. -1 = random. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | — |