Skip to content

A2M v3.0.0

Choose a tag to compare

@Justagwas Justagwas released this 14 Jul 17:45
· 5 commits to main since this release
6e9151e

Release Notes for Version: 2.0.2 → 3.0.0

Included Files

  • A2MSetup.exe (Recommended)
    Windows installer for A2M. Size: 69.2 MB
    SHA-256: 898c575338ee010597d1224dc5da8f3db9f35191f5ed1c91752f52d55b1211c9

  • A2M.exe
    Standalone A2M application. Size: 99.5 MB
    SHA-256: ae6d77b0c24247cd852b07182bf832baafbebc3c238089499b7db9c6532ce8c3

  • Source code (zip / tar.gz)
    Compressed A2M source code.


Security Notice

Only download from these sources: Official repo ; Sourceforge ; Website.


Changes

Major

  • Replaced the Legacy and Modern transcription backends with a newer accuracy-focused A2M Piano Engine.
  • Introduced a new multi-component Transcription Model package.
  • Redesigned GPU acceleration around explicit CUDA and DirectML choices.
  • Added separate, system-aware CPU usage and GPU memory controls.
  • Added a first-launch initialization screen that sets A2M ready for usage.
  • Added a complete v1/v2 upgrade path that migrates existing settings, preserves exported MIDI files, and removes retired models, incompatible runtime packs, and obsolete backend data.

Minor

  • Added controls for pedal output, expressive or uniform note velocity, and uniform velocity strength.
  • Added independent CPU usage settings for CPU and GPU transcription.
  • A2M now reduces GPU processing groups automatically when a selected memory level is unsupported.
  • GPU setup now returns to the user's intended GPU selection after installing a required runtime pack, while failed provider changes restore the previous working configuration.
  • Added a post-conversion MIDI summary showing note count, duration, pitch range, pedal count, processing time, and save location.
  • Improved CUDA and DirectML status messages and disabled GPU provider controls while CPU mode is selected.
  • Improved cancellation and lower-memory audio decoding for large or less commonly supported audio files.
  • Strengthened model, runtime-pack, and update downloads with size, source, version, and SHA-256 validation.
  • Improved settings migration and saving to reduce the chance of lost or partially written preferences.

Full Changelog: v2.0.2...v3.0.0