v1.3.3

ByteDance-Seed/SeedVRv1.3.3May 23, 2026by LauraGPT

AI Summary

This release adds agent integration capabilities to FunASR, including a new CLI server, an MCP server for AI assistants, and an OpenAI-compatible API endpoint for speech transcription.

Key Highlights

New `funasr-server` CLI for starting an OpenAI-compatible transcription server
New MCP Server for AI assistants (Claude, Cursor, Windsurf) to transcribe audio
OpenAI-Compatible API works with LangChain, AutoGen, and other frameworks
Bug fixes for HuggingFace Hub parameter propagation and Qwen3-ASR ImportError

New Features

funasr-server CLI tool
MCP Server integration
OpenAI-compatible `/v1/audio/transcriptions` endpoint

Full Release Notes

## Highlights

This release makes FunASR a drop-in speech backend for AI agents.

### New: `funasr-server` CLI
```bash
pip install funasr fastapi uvicorn python-multipart
funasr-server --device cuda
```
One command starts an OpenAI-compatible `/v1/audio/transcriptions` endpoint.

### New: MCP Server
AI assistants (Claude, Cursor, Windsurf) can now transcribe audio directly.

### New: OpenAI-Compatible API
Works with **any** agent framework: LangChain, AutoGen, CrewAI, Dify, Flowise, Open WebUI.

```python
from openai import OpenAI
client = OpenAI(base_url="http://localhost:8000/v1", api_key="x")
result = client.audio.transcriptions.create(model="sensevoice", file=open("a.wav","rb"))
```

### Bug Fixes
- Fixed `hub="hf"` parameter propagation to sub-models (v1.3.2)
- Fixed Qwen3-ASR ImportError masking

### Upgrade
```bash
pip install --upgrade funasr
```

### Links
- [Agent Integration Guide](https://modelscope.github.io/FunASR/agent.html)
- [OpenAI API Docs](https://github.com/modelscope/FunASR/tree/main/examples/openai_api)
- [MCP Server Docs](https://github.com/modelscope/FunASR/tree/main/examples/mcp_server)
- [Benchmark](https://modelscope.github.io/FunASR/benchmark.html)