ComfyUI Node
T5Gemma-TTS Generate
Generate speech from text using T5Gemma-TTS. Supports multilingual TTS (EN/JA/ZH), voice cloning, and duration control.
T5Gemma-TTS Generate
- model
- reference_audio
- audio
◄target_textHello, this is a test of the text to speech system.►
◄top_k30►
◄top_p0.90►
◄temperature0.80►
◄seed1►
◄reference_text►
◄target_duration0.0►
◄languageauto►
Categoryaudio/tts
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| model | T5GEMMA_MODEL | T5Gemma-TTS model from the ModelLoader node | |
| target_text | STRING | Hello, this is a test of the text to speech system. | Text to synthesize into speech |
| top_k | INT | 301–200 | Top-k sampling parameter |
| top_p | FLOAT | 0.900–1 | Top-p (nucleus) sampling parameter |
| temperature | FLOAT | 0.800.01–2 | Sampling temperature — lower = more deterministic |
| seed | INT | 10–2147483647 | Random seed for reproducibility |
| reference_audioopt | AUDIO | Reference audio for voice cloning (optional) | |
| reference_textopt | STRING | Transcript of reference audio (leave blank to auto-transcribe via Whisper) | |
| target_durationopt | FLOAT | 0.00–120 | Target audio duration in seconds (0 = auto-estimate from text) |
| languageopt | COMBO | auto | Language of the target text (auto = detect automatically) |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | — |