pythonfix
funasr 1.3.1
FunASR: A Fundamental End-to-End Speech Recognition Toolkit
FunASR: A Fundamental End-to-End Speech Recognition Toolkit
Stars: 14915, Watchers: 14915, Forks: 1577, Open Issues: 538The modelscope/FunASR repo was created 3 years ago and the last code push was 2 weeks ago.
The project is extremely popular with a mindblowing 14915 github stars!
How to Install funasr
You can install funasr using pip
pip install funasr
or add it to a project with poetry
poetry add funasr
Package Details
- Author
- Speech Lab of Alibaba Group
- License
- The MIT License
- Homepage
- https://github.com/alibaba-damo-academy/FunASR.git
- PyPi:
- https://pypi.org/project/funasr/
- GitHub Repo:
- https://github.com/alibaba-damo-academy/FunASR
Classifiers
- Software Development/Libraries/Python Modules
Related Packages
Errors
A list of common funasr errors.
Code Examples
Here are some funasr code examples and snippets.
GitHub Issues
The funasr package has 538 open issues on GitHub
- Codex/create training and inference scripts
- 怎么查看docker的镜像发布到什么版本了？
- 使用CAM++识别多个说话人时，报错AssertionError。
- 请问一下，实时转文字都是使用的cpu版本，是cpu的比gpu的快吗？有没有服务端的代码。我看全是docker镜像，有没有分支代码可以参考
- model.generate时，传入的是音频字节流，会对采样率校验耗时200ms。
- 咨询模型最长支持的音频时长
- paraformer_v2_community offline
- fa-zh 强制对齐模型时间戳存在系统性偏移
- 钉钉交流群无法加入
- SenseVoice output_timestamp=True, # 必须显式打开再用vad模型报错
- GPU镜像拉下之后跑会报错，尝试转onnx发现跟外部CUDA12.2不兼容，里面构建的CUDA太老。
- Non-default 'decibel_thres' value causes FSMN VAD model crash
- 修复流式识别中prev_samples截取bug
- 测试100次前30次3分钟音频，后70次50秒，内存增加到1.95GiB，使用funasr-runtime-sdk-cpu-0.4.7 ，推理结束内存一直不释放
- speech_whisper-large-v3/model_quant.onnx does not exist.
Related Packages & Articles
speechbrain 1.0.3
All-in-one speech toolkit in pure Python and Pytorch
paddlespeech 1.5.0
Speech tools and models based on Paddlepaddle
transformers 5.2.0
Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Adeepspeed 0.9.2
DeepSpeed library
auto-gptq 0.7.1
An easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.
mmdet 3.3.0
OpenMMLab Detection Toolbox and Benchmark
neuralforecast 3.1.4
Time series forecasting suite using deep learning models
detectors 0.1.11
This is a comprehensive library for generalized Out-of-Distribution (OOD) detection research. It provides over 20 detection methods, evaluation pipelines, OOD datasets, and model architectures integrated with timm. Additionally, it offers fast OOD evaluation metrics, multi-layer detection methods, and pipelines for open set recognition and covariate drift detection.
openai-whisper 20250625
Robust Speech Recognition via Large-Scale Weak Supervision