ComfyUI Node
T5Gemma-TTS Model Loader
Load a T5Gemma-TTS model and tokenizers for text-to-speech generation.
T5Gemma-TTS Model Loader
- model
◄model_dirAratako/T5Gemma-TTS-2b-2b►
◄deviceauto►
◄xcodec2_model_name►
◄use_torch_compilefalse►
◄cpu_codecfalse►
◄cpu_whisperfalse►
Categoryaudio/tts
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| model_dir | STRING | Aratako/T5Gemma-TTS-2b-2b | HuggingFace repo ID or local path to the model |
| device | COMBO | auto | Device for model inference |
| xcodec2_model_nameopt | STRING | Override XCodec2 model name (blank = use config default) | |
| use_torch_compileopt | BOOLEAN | false | Apply torch.compile for faster repeat inference (CUDA only) |
| cpu_codecopt | BOOLEAN | false | Run XCodec2 audio codec on CPU to save ~3.5 GB VRAM |
| cpu_whisperopt | BOOLEAN | false | Run Whisper transcription on CPU to save ~5 GB VRAM |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | T5GEMMA_MODEL | — |