search

k2-fsa / sherpa-onnx

Speech-to-text, text-to-speech, speaker diarization, and VAD using next-gen Kaldi with onnxruntime without Internet connection. Support embedded systems, Android, iOS, Raspberry Pi, RISC-V, x86_64 servers, websocket server/client, C/C++, Python, Kotlin, C#, Go, NodeJS, Java, Swift, Dart, JavaScript, Flutter, Object Pascal, Lazarus, Rust
https://k2-fsa.github.io/sherpa/onnx/index.html
Apache License 2.0
3.59k stars 423 forks source link

Shallow fusion with N-gram Language Model #182

Open tienanh28122000 opened 1 year ago

tienanh28122000 commented 1 year ago

Hi every one, do we have a shallow fusion implementation with N-gram Language Model (e.g Kenlm or Srilm) instead of Neural Language Model? If not, can you give me some instructions to do shallow fusion with N-gram and then maybe I can have a PR for this. Thanks in advance!

marcoyang1998 commented 1 year ago

Hi, n-gram LM is usually much weaker than a neural network LM (such as RNN/Transformer). We only supported RNNLM rescoring, see #125.