A2M (Audio To MIDI)

Desktop audio-to-MIDI transcription with optional CUDA/DirectML GPU acceleration

Overview

A2M converts piano-focused audio into MIDI locally on your Windows PC. Choose an audio file, run transcription, and export a .mid file for review or editing.

Best results on solo piano takes and clean stems
Local transcription with no cloud upload
CPU default, with optional CUDA/DirectML GPU acceleration
Legacy and Modern transcription engines (Legacy default)
Modern tuning controls for thresholds, denoise, overlap stitching, and auto calibration
Stop support for model download, runtime install, and conversion
MIDI output saved to Downloads/A2M

Download

Version 2.0.1 - February 20, 2026

Download (Official)

GitHub SourceForge

The installer includes everything you need to get started. It will guide you through installation and install the required components for A2M.

Downloads this week

Prefer browsing first? See the Preview.

Preview

Real UI captures from A2M.

Latest release and safety

Version 2.0.1 - February 20, 2026

Updated main window with a new pin toggle.
Migrated constants to config.
Improved update safety, cursor handling.
Slight performance increase.
Minor other UI fixes.

Security and installation

A2M is safe. Windows may still show a "Protected your PC" warning because the app is new or unsigned. Choose More info and then Run anyway if prompted.

The latest installer was scanned on VirusTotal with: 0/70 detections. View VirusTotal scan.

Manual installer and release details are available on the download page.

Open source and distribution

A2M is open source and available on both GitHub and SourceForge.
The transcription model is based on piano_transcription_inference.

GitHub SourceForge Releases License