flutter_whisper 0.2.0

On-device speech-to-text transcription using whisper.cpp. Automatic model download, streaming segment results, Android support.

0.2.0

• Subtitle Exporters: Added result.toSrt() (SubRip) and result.toVtt() (WebVTT) formatters with millisecond timestamp precision.
• In-Memory Audio Transcription: Added Whisper.transcribeBytes() for transcribing Uint8List byte buffers and raw PCM data directly.
• Audio Utilities: Added AudioUtils for 44-byte RIFF WAV header generation, PCM-to-WAV conversion, WAV duration calculation, and header validation.
• Drop-in UI Widgets:
  • WhisperRecordingButton: Animated pulsing microphone button with timer and start/stop/cancel actions.
  • TranscriptionView: Material 3 transcription viewer with timestamp chips, word pills, copy-to-clipboard, and segment navigation.
• Model Specifications: Added WhisperModel metadata for RAM usage estimates, speed/accuracy ratings, file sizes, and recommended use-cases.
• Expanded Test Suite: Added comprehensive unit and widget tests covering subtitle formatters, audio utilities, model specs, and UI widgets.

0.1.0

• Initial release: on-device speech-to-text using whisper.cpp (Android).
• Automatic model download from HuggingFace (tiny/base/small/medium/large).
• Native whisper.cpp engine via JNI bridge, CPU inference (ARM64 / ARMv7).
• Streaming segment results for real-time UI updates.
• Segment + word-level timestamps, language auto-detection and forcing.