Piano Transcriptionist

New Transcription Model This update introduces a new default transcription model based on Transkun. (#6) Built on a transformer architecture, the same family of models used in modern AI chat systems, it provides better context awareness and improved transcription accuracy. The transcription pipeline has also been optimized with parallel CPU and GPU processing. Because the new model is more computationally complex, it may still run slower than the previous model on some devices. You can switch between transcription models at any time in Settings. Bug Fixes and Improvements • Improved the bug report view on iOS. • Transcription tasks are now properly cancelled when closing the transcription window on macOS. • Improved localized window names on macOS. • Improved efficiency when transcribing multiple files at once. • Fixed a sustain-related bug when importing MIDI files. • Improved the audio normalization algorithm. The previous version could be too aggressive and sometimes caused notes to be cut short.