KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#developer-tools #dsh #dsh-plugin #ai-agent #claude-code #codex #typescript #ai-agents #mcp #cli #rust #deepseek-harness

ComfyUI-Qwen3-ASR

★ 27 Open GitHub ↗

ComfyUI nodes for Qwen3-ASR (0.6B/1.7B) and ForcedAligner. Supports high-accuracy ASR and language identification for 52 languages/dialects, including 22 Chinese dialects and various English accents. Features word-level timestamps, long audio transcription, and VRAM-optimized inference.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# ComfyUI Qwen3 ASR

A high-performance ComfyUI integration for the **Qwen3-ASR** model family. This extension provides state-of-the-art speech-to-text transcription, language identification, and precise word-level timestamps using the novel Qwen3 Forced Aligner.

[![Model](https://img.shields.io/badge/Model-Qwen3--ASR-blue)](https://github.com/QwenLM/Qwen3-ASR)
[![License](https://img.shields.io/badge/License-MIT-green)](https://raw.githubusercontent.com/kaushiknishchay/ComfyUI-Qwen3-ASR/main/LICENSE.md)

## Features
- **High Accuracy**: Supports Qwen3-ASR 0.6B and 1.7B models.
- **Multilingual**: Supports 52 languages and dialects with automatic language detection.
- **Word-Level Timestamps**: Optional integration with Qwen3-ForcedAligner-0.6B.
- **Flexible Precision**: Support for `bf16`, `fp16`, and `fp32` to balance VRAM and speed.
- **Automatic Resampling**: Internally handles audio resampling to 16kHz for optimal model performance.
- **FlashAttention 2**: Integrated support for FlashAttention 2 to significantly reduce VRAM usage and accelerate inference.

## Preview

![Preview](https://raw.githubusercontent.com/kaushiknishchay/ComfyUI-Qwen3-ASR/main/assets/image.png)

## Installation

### Manual Installation

1. Navigate to your ComfyUI `custom_nodes` directory:
   ```bash
   cd ComfyUI/custom_nodes
   ```
2. Clone this repository:
   ```bash
   git clone https://github.com/kaushiknishchay/ComfyUI-Qwen3-ASR
   ```
3. Install the dependencies using your ComfyUI Python executable:
   ```bash
   # For portable versions, use the full path to your python.exe
   python.exe -m pip install -r ComfyUI-Qwen3-ASR/requirements.txt
   ```
4. **(Recommended)** Install FlashAttention 2 for maximum performance:
   ```bash
   # For FlashAttention 2 (requires compatible NVIDIA GPU)
   python.exe -m pip install -U flash-attn --no-build-isolation
   ```

### Install via ComfyUI Manager
- Search `ComfyUI-Qwen3-ASR` by `Kaushik`

## Model Setup

Models must be placed in the `models/diffusion_models/Qwen3-ASR/` directory. Each model should be in its own subfolder containing the full weights and configuration.

### Recommended Directory Structure:
```
ComfyUI/models/diffusion_models/Qwen3-ASR/
├── Qwen3-ASR-1.7B/
│   ├── config.json
│   ├── model.safetensors
│   └── ...
├── Qwen3-ASR-0.6B/
│   └── ...
└── Qwen3-ForcedAligner-0.6B/
    └── ...
```

### Downloading Models
You can use the `huggingface-cli` to download the models directly into the correct folders:
```bash
huggingface-cli download Qwen/Qwen3-ASR-1.7B --local-dir models/diffusion_models/Qwen3-ASR/Qwen3-ASR-1.7B
huggingface-cli download Qwen/Qwen3-ForcedAligner-0.6B --local-dir models/diffusion_models/Qwen3-ASR/Qwen3-ForcedAligner-0.6B
```

## Node Usage

### 1. Qwen3 ASR Transcriber
The main node for generating transcriptions.

**Inputs:**
- `audio`: The audio input from a "Load Audio" node.
- `model_name`: Select the ASR model folder from the dropdown.
- `language`: Set to `auto` for automatic detection, or specify a language (e.g., `English`, `Chinese`).
- `device`: `cuda` (recommended) or `cpu`.
- `precision`: `bf16` (recommended for RTX 30/40 series), `fp16`, or `fp32`.
- `max_new_tokens`: Maximum length of the generated text.
- `flash_attention_2`: Enable Flash Attention 2 for faster inference and lower VRAM usage.
- `chunk_size`: Process audio in chunks of this many seconds (default: 30). Set to 0 to disable.
- `overlap`: Overlap between chunks in seconds to maintain context (default: 2).
- `forced_aligner` (Optional): Connect the output of a **Qwen3 Forced Aligner Config** node here to enable timestamps.

**Outputs:**
- `text`: The raw transcription text.
- `timestamps`: A formatted string containing word-level timestamps (requires the Forced Aligner).

### 2. Qwen3 Forced Aligner Config
A configuration node that prepares the alignment model.

**Inputs:**
- `model_name`: Select the `Qwen3-ForcedAligner-0.6B` folder.
- `device`: Should match the Transcriber node.
- `precision`: Should match the Transcriber node.
- `flash_attention_2`: Enable Flash Attention 2 for the aligner model to improve performance and reduce memory footprint.

**Outputs:**
- `aligner_config`: Connect this to the `forced_aligner` input on the Transcriber node.

## Troubleshooting

- **Python 3.13 Issues**: If you encounter an `UnboundLocalError` related to `lazy_loader`, ensure you have updated the package:
  ```bash
  python.exe -m pip install -U lazy-loader
  ```
- **VRAM Usage**: The 1.7B model requires approximately 4-6GB of VRAM in `bf16` mode. If you run out of memory, try the 0.6B model or use `cpu` mode.

## License
This project is licensed under the MIT License. The Qwen3 models are subject to the Qwen License Agreement.

ComfyUI-Qwen3-ASR

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README