29 releases (4 breaking)

0.5.1 Oct 25, 2024
0.5.1-beta.0 Oct 19, 2024
0.4.1 Oct 18, 2024
0.3.0 Oct 16, 2024
0.1.7-beta.0 Jul 13, 2024

#159 in Audio

Download history 8/week @ 2024-08-19 14/week @ 2024-08-26 39/week @ 2024-09-16 128/week @ 2024-09-30 2/week @ 2024-10-07 794/week @ 2024-10-14 220/week @ 2024-10-21 19/week @ 2024-10-28 11/week @ 2024-11-04 7/week @ 2024-11-18 380/week @ 2024-12-02

388 downloads per month

MIT license

2MB
52K SLoC

C++ 29K SLoC // 0.1% comments Kotlin 8K SLoC // 0.1% comments Java 6K SLoC // 0.0% comments Dart 3K SLoC // 0.0% comments C 2K SLoC // 0.2% comments Rust 2K SLoC // 0.0% comments Batch 1K SLoC Prolog 269 SLoC Python 149 SLoC // 0.1% comments PowerShell 75 SLoC // 0.1% comments JavaScript 72 SLoC Shell 37 SLoC Forge Config 1 SLoC

sherpa-rs

Crates License

Rust bindings to sherpa-onnx

Features

  • Spoken language detection
  • Speaker embedding (labeling)
  • Speaker diarization
  • Speech to text
  • Text to speech
  • Text punctuation
  • Voice activity detection
  • Audio tagging
  • Keyword spotting

Supported Platforms

  • Windows
  • Linux
  • macOS

Install

cargo add sherpa-rs

Build

Please see BUILDING.md.

Feature flags

  • cuda: enable CUDA support
  • directml: enable DirectML support
  • tts: enable TTS
  • download-binaries: use prebuilt sherpa-onnx libraries for faster builds. cached.
  • static: use static sherpa-onnx libraries and link them statically.
  • sys: expose raw c bindings (sys crate)

Docs

See sherpa/intro.html

Examples

See examples

Models

All pretrained models available at sherpa/onnx/pretrained_models

Dependencies

~0.6–4MB
~68K SLoC