sherpa-onnx Releases

The release primarily targets Android packaging with the provision of the sherpa-onnx AAR file and updates speaker identification support for HarmonyOS. It also includes a version fix for portaudio-go.

Dec 12, 2024

This release significantly expands the HarmonyOS ecosystem by adding on-device real-time ASR, speaker identification, and speaker diarization capabilities. It also includes various documentation updates and fixes for node examples.

Dec 10, 2024

A major update introducing full streaming ASR and text-to-speech support for the HarmonyOS platform, along with VAD+ASR demos and microphone examples. It also publishes the necessary HAR packages for the platform.

Dec 4, 2024

This release enables cross-compilation for HarmonyOS and adds VAD support to the platform, alongside a Flutter iOS build fix.

Nov 26, 2024

This release adds support for Python 3.13 and GPU-capable builds on Linux aarch64, while expanding examples for Moonshine and WebAssembly models. It also introduces static builds for Windows ARM64.

Nov 16, 2024

Adds support for the Moonshine model with a wide range of language APIs and fixes Windows node-addon build issues.

Oct 27, 2024

Introduces Russian ASR support via GigaAM models and speaker diarization capabilities, alongside various API updates.

Oct 25, 2024

Adds support for Parakeet and Whisper Turbo models, plus comprehensive speaker diarization APIs across multiple programming languages.

Oct 13, 2024

Minimal release notes: "Please see https://k2-fsa.github.io/sherpa/onnx/speaker-diarization/index.html"

Sep 29, 2024

Fixes build issues and adds non-streaming Russian ASR models.

Sep 19, 2024

Enhances SenseVoice support and adds configuration options for VAD speech duration limits.

Sep 14, 2024

Improves stability with error handling and adds embedded system support, alongside online punctuation model bindings.

Sep 13, 2024

Aug 30, 2024

Adds Object Pascal support and enhances mobile platform compatibility with VAD and TTS features.

Aug 24, 2024

This release introduces Pascal API support for streaming and non-streaming ASR, VAD, and wave file reading. It also adds support for multi-channel wave files and SenseVoice emotion detection, along with various examples for subtitle generation and asset management.

Aug 16, 2024

The update focuses on enhancing Java and Flutter support, including a non-streaming WebSocket client for Java and a Chinese+English TTS example. It also adds a Japanese pre-trained model for ReazonSpeech and an English online punctuation and casing prediction model.

Aug 8, 2024

This release adds a TTS example for the Java API and introduces new Dart API features for audio tagging and text punctuation. It also includes examples combining VAD with non-streaming ASR.

Jul 29, 2024

A significant refactoring of the C API to ensure consistent naming conventions by prefixing all functions with 'SherpaOnnx'.

Jul 26, 2024

Added a JavaScript API example demonstrating the integration of Voice Activity Detection (VAD) with non-streaming Automatic Speech Recognition (ASR).

Jul 26, 2024

Extensive update adding SenseVoice support across multiple languages including C++, C#, Go, JavaScript, WebAssembly, Dart, and Java/Kotlin. Also introduces DirectML support and exports SenseVoice to ONNX.

Jul 23, 2024