Skip to content

CHANGELOG.md

Latest commit

 

History

History
49 lines (42 loc) · 2.23 KB

CHANGELOG.md

File metadata and controls

49 lines (42 loc) · 2.23 KB

Changelog

All notable changes to the DuckDB Whisper Extension will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.1.0] - 2025-01-28

Added

  • Initial release of the DuckDB Whisper Extension
  • Core transcription functions:
    • whisper_transcribe() - Transcribe audio to text (scalar function)
    • whisper_transcribe_segments() - Transcribe with timestamps and metadata (table function)
  • Model management functions:
    • whisper_list_models() - List all available models with download status
    • whisper_download_model() - Get download instructions for a model
  • Utility functions:
    • whisper_version() - Get extension and whisper.cpp version info
    • whisper_check_audio() - Validate audio files
    • whisper_audio_info() - Get audio file metadata
  • Configuration functions (session-persistent):
    • whisper_set_model() - Set default model
    • whisper_set_model_path() - Set model storage directory
    • whisper_set_language() - Set default language
    • whisper_set_threads() - Set thread count
    • whisper_get_config() - View current configuration
  • Recording functions (requires SDL2, enabled by default):
    • whisper_list_devices() - List audio input devices
    • whisper_record() - Record from microphone and transcribe
    • whisper_record_translate() - Record and translate to English
    • whisper_record_auto() - Record with automatic silence detection (configurable)
  • Support for all Whisper models (tiny through large-v3-turbo)
  • Support for multiple audio formats via FFmpeg (WAV, MP3, FLAC, OGG, AAC, etc.)
  • BLOB input support for transcribing audio from memory
  • Automatic audio conversion to 16kHz mono (Whisper requirement)
  • Model caching for improved performance on repeated transcriptions
  • SQL test suite with 34 test assertions
  • Comprehensive documentation
  • Compile-time option WHISPER_ENABLE_RECORDING (ON by default)

Technical Details

  • Built on whisper.cpp v1.8.3
  • Compatible with DuckDB v1.4.x
  • Uses FFmpeg for audio decoding
  • Uses SDL2 for audio recording (optional)
  • Supports macOS (ARM64, x86_64) and Linux