yeyupiaoling/MASR
A Pytorch automatic speech recognition framework supporting streaming and non-streaming ASR with Conformer, Squeezeformer, and DeepSpeech2 models.
Not currently ranked — collecting fresh signals.
star history
MASR is a speech recognition framework implementing CTC and attention-based decoding strategies including greedy search, prefix beam search, and attention rescoring. It supports multiple preprocessing features like fbank and mfcc, and various data augmentation techniques. The project targets deployment on servers, Nvidia Jetson devices, and plans to support mobile platforms.
Frequently asked
- What is yeyupiaoling/MASR?
- A Pytorch automatic speech recognition framework supporting streaming and non-streaming ASR with Conformer, Squeezeformer, and DeepSpeech2 models.
- Is MASR open source?
- Yes — yeyupiaoling/MASR is open source, released under the Apache-2.0 license.
- What language is MASR written in?
- yeyupiaoling/MASR is primarily written in Python.
- How popular is MASR?
- yeyupiaoling/MASR has 728 stars on GitHub.
- Where can I find MASR?
- yeyupiaoling/MASR is on GitHub at https://github.com/yeyupiaoling/MASR.