Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
A ComfyUI custom node suite for Qwen3-TTS, supporting 1.7B and 0.6B models, Custom Voice, Voice Design, Voice Cloning and Fine-Tuning.
| Date | Stars |
|---|---|
| 2026-07-24 | 288 |
| 2026-07-25 | 288 |
| 2026-07-28 | 288 |
| 2026-07-30 | 289 |
| 2026-08-06 | 289 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# ComfyUI Qwen3-TTS
A ComfyUI custom node suite for [Qwen3-TTS](https://github.com/QwenLM/Qwen3-TTS), supporting 1.7B and 0.6B models, Custom Voice, Voice Design, Voice Cloning and Fine-Tuning.
> 🎤 **Looking for Speech-to-Text?** Check out [ComfyUI-Qwen3-ASR](https://github.com/DarioFT/ComfyUI-Qwen3-ASR) for audio transcription with compatible outputs!
<p align="center">
<img src="https://raw.githubusercontent.com/DarioFT/ComfyUI-Qwen3-TTS/refs/heads/main/assets/intro.png"/>
<p>
## Features
- **ComfyUI Model Folder Integration**: Models are stored in `ComfyUI/models/Qwen3-TTS/`, keeping your models organized alongside other ComfyUI models.
- **On-Demand Download**: Only downloads the model you select—no need to pre-download all variants.
- **Full Qwen3-TTS Support**:
- **Custom Voice**: Use 9 preset high-quality voices (Vivian, Ryan, etc.).
- **Voice Design**: Create new voices using natural language descriptions.
- **Voice Cloning**: Clone voices from a short reference audio clip.
- **Fine-Tuning**: Train a custom voice model using your own dataset (folder of .wav + .txt files).
- Resume training from checkpoints
- VRAM optimizations: gradient checkpointing, 8-bit AdamW, configurable batch sizes
- Per-epoch checkpointing with automatic cleanup
- Support for both 1.7B and 0.6B models
- **Audio Comparison**: Evaluate fine-tuned models with speaker similarity and mel spectrogram metrics.
- **Cross-Lingual Support**: Generate speech in Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, and Italian.
- **Flexible Attention**: robust support for `flash_attention_2` with automatic fallback to `sdpa` (standard PyTorch 2.0 attention) if dependencies are missing.
## Installation
1. Clone this repository into your `ComfyUI/custom_nodes` folder:
```bash
cd ComfyUI/custom_nodes
git clone https://github.com/DarioFT/ComfyUI-Qwen3-TTS.git
```
2. Install dependencies:
```bash
cd ComfyUI-Qwen3-TTS
pip install -r requirements.txt
```
**For portable/standalone ComfyUI installations**, use the embedded Python instead:
```bash
# From your ComfyUI_windows_portable root folder
.\python_embeded\python.exe -m pip install -r ComfyUI\custom_nodes\ComfyUI-Qwen3-TTS\requirements.txt
```
> ⚠️ **Note**: ComfyUI does not auto-install dependencies from `requirements.txt`. You must run the install command manually.
*For GPU acceleration, ensure you have a CUDA-compatible PyTorch installed.*
> ⚠️ **Dependency Note**: The upstream `qwen-tts` package requires `transformers==4.57.3`. This may downgrade your existing transformers version. If other custom nodes require a newer version, consider using a separate Python environment.
## Model Storage
Models and tokenizers are automatically stored in your ComfyUI models folder:
```
ComfyUI/models/Qwen3-TTS/
├── Qwen3-TTS-12Hz-1.7B-CustomVoice/
├── Qwen3-TTS-12Hz-1.7B-VoiceDesign/
├── Qwen3-TTS-12Hz-1.7B-Base/
├── Qwen3-TTS-12Hz-0.6B-CustomVoice/
├── Qwen3-TTS-12Hz-0.6B-Base/
├── Qwen3-TTS-Tokenizer-12Hz/ # For fine-tuning
└── prompts/ # Saved voice embeddings (.safetensors)
```
**First-time use**: When you first select a model and run the workflow, it will be downloaded automatically. Only the model you select is downloaded—not all variants.
**Existing cached models**: If you previously used this extension and have models in HuggingFace (`~/.cache/huggingface/hub/`) or ModelScope (`~/.cache/modelscope/hub/`) cache, they will be automatically migrated to the ComfyUI models folder.
## Usage
### 1. Load Model
Use the **Qwen3-TTS Loader** node.
- **repo_id**: Select the model you want to use.
- `CustomVoice` models: For using preset speakers.
- `VoiceDesign` models: For designing voices with text prompts.
- `Base` models: For voice cloning and fine-tuning.
- **source**: Choose between HuggingFace or ModelScope for downloading (if model not alreExcerpt of 9,432 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:c39fbacd1ccaac8d, topic:voice-cloning, readme:speech-to-text, desc:voice cloning
matched fp:c39fbacd1ccaac8d, desc:fine-tuning, readme:fine-tuning, desc:fine tuning
matched fp:c39fbacd1ccaac8d, topic:comfyui