asr_lib 0.6.0 copy "asr_lib: ^0.6.0" to clipboard
asr_lib: ^0.6.0 copied to clipboard

On-device speech recognition for Flutter powered by sherpa-onnx.

0.6.0 #

  • bump version

0.5.1 #

  • Align funasrNano with senseVoice config and model layout.

0.5.0 #

  • Version bump to match pubspec.yaml.

0.3.0 #

Breaking change: This release removes all Android/Kotlin native code and replaces the Platform Channel bridge with a pure-Dart implementation using the sherpa_onnx Flutter package.

  • Pure-Dart ASR engineAsrLib now uses sherpa_onnx directly instead of the previous Kotlin native AAR + Platform Channel bridge.
  • Cross-platform support — works on Android, iOS, macOS, Windows, and Linux (all platforms supported by sherpa_onnx + record).
  • Audio capture via record package — replaces Android AudioRecord.
  • Model downloader in Dart — new asr_model_downloader.dart replaces the Kotlin ModelDownloader. Supports progress tracking and marker-file-based download state.
  • Bundled VAD modelsilero_vad.onnx is now a Flutter asset.
  • android/ directory removed — no native code needed.
  • Model type supportAsrConfig.modelType selects the model architecture: zipformer_ctc (default), sense_voice, or paraformer.
  • New dependencies: record, path_provider, path.
  • plugin: declaration removed from pubspec.yaml.

0.2.0 #

  • Permission handlingAsrPermissions for microphone permission request/check (Dart layer).
  • Global service registryAsrRegistry singleton with pre-initialization support.
  • UI componentsVoiceRecordButton (long-press recording) and showVoiceRecordOverlay (full-screen recording dialog).
  • Animation widgetsPulseAnimation, RippleAnimation, SoundLevelIndicator, WaveformDisplay.
  • Text correctionGlobalTextCorrector with pinyin matching, fuzzy matching, and hotword support.
  • New dependencies: permission_handler, lpinyin.

0.1.0 #

  • Initial release.
  • Android arm64-v8a support (minSdk 24).
  • Streaming (ONLINE) and full-utterance (OFFLINE) ASR modes.
  • Silero VAD for automatic speech segment detection.
  • Built-in mic capture via Android AudioRecord.
  • Model auto-download from ModelScope (funasr-nano).
  • Flutter plugin with Platform Channel bridge (MethodChannel + EventChannel).
  • Kotlin native AAR for use outside Flutter.
0
likes
150
points
217
downloads

Documentation

API reference

Publisher

unverified uploader

Weekly Downloads

On-device speech recognition for Flutter powered by sherpa-onnx.

Homepage

License

Apache-2.0 (license)

Dependencies

dio, flutter, path, path_provider, permission_handler, record, sherpa_onnx

More

Packages that depend on asr_lib