# Audio to MIDI AudioPod converts songs and stems into editable MIDI on every plan, Basic included: upload a track, AudioPod's audio-to-MIDI engine separates it and transcribes bass, vocals, piano and guitar (experimental) into one .mid file per instrument plus a merged multitrack file, at 960 credits per minute of audio. Canonical URL: https://audiopod.ai/features/audio-to-midi Last updated: 2026-10-14 ## Facts | Fact | Detail | | --- | --- | | Instruments transcribed | Bass, vocals, piano and guitar (experimental) | | Selected by default | Bass, vocals and piano | | Not transcribed | Drums, percussion and the leftover "other" part | | Input | An audio or video file, or a link; at least 6 seconds long | | Output | One .mid per instrument, one merged multitrack .mid, and the separated stem audio | | Tempo | Detected automatically by beat tracking, or set 30–300 BPM through the API | | From a full song | 960 credits per minute (the stem split is included) | | From a stem you already split | 300 credits per minute of the stem | | Minimum charge | 10 credits | | Piano / vocal specialists (Creator+) | 990 credits per minute converting an existing stem; 1,728 per minute from a full song (split included) | | API | POST /api/v1/midi/convert · POST /api/v1/midi/convert/from-stems/{stem_job_id} · MCP tool audio_to_midi | ## How to do it 1. Open Audio to MIDI and upload a song or paste a link. 2. Pick the instruments to transcribe (bass, vocals and piano are on by default). 3. Check the credit estimate and run the job. 4. Download the per-instrument .mid files or the merged multitrack file and open them in your DAW. ## How a conversion runs Each job moves through 3 stages, shown live in the tool: Separating stems, Transcribing notes and Merging MIDI tracks. - Separating stems: the song is split so each instrument is transcribed on its own, not from the full mix. - Transcribing notes: each selected instrument is turned into notes with onsets, lengths and pitches. - Merging MIDI tracks: the per-instrument files are combined into one multitrack .mid with the tempo and a named instrument on every track (vocals as Lead 6 (voice), bass as Electric Bass (finger), guitar as Electric Guitar (clean) and piano as Acoustic Grand Piano). ## What to expect MIDI is a starting point — expect to tidy timing and note lengths. Drums aren't transcribed yet. Dynamics are approximate. Works best on clean recordings. - Bass: Transcribes most reliably — the best place to start. - Vocals: Lead melody line. - Piano: Chords and melody. - Guitar (experimental, off by default): Dense or heavily-strummed guitar is hit-or-miss. - Drums aren't transcribed yet — coming later. ## Tuning through the API API callers can override the note-detection defaults per job: - onset_threshold (default 0.5) — how confident a note start must be. - frame_threshold (default 0.3) — how confident a sustained note must be. - min_note_len_ms (default 128) — the shortest note kept. - tempo_bpm (30–300) — skip beat tracking and write this tempo. - engine — general, or the piano or vocal specialist on Creator and above. ## FAQ **Which instruments can AudioPod convert to MIDI?** Bass, vocals, piano and guitar (experimental). Drums aren't transcribed yet — coming later. **How much does audio to MIDI cost?** 960 credits per minute from a full song, the split included — a 4-minute song is 3,840 credits (about $0.51 pay-as-you-go). Converting a stem you already split costs 300 credits per minute of that stem, 1,200 for 4 minutes. Every plan, including the free Basic plan, can run it from its credits. **Can I convert a stem I already split?** Yes. After a Stem Splitter job, press Convert to MIDI under a bass, vocals, piano or guitar stem. That costs 300 credits per minute of the stem instead of the 960 full-song rate, because the split is already done. **What files do I get?** One .mid file for each instrument, a merged multitrack .mid with the tempo and a named instrument on each track, and, when you start from a full song, the separated stem audio. **How accurate is the MIDI?** MIDI is a starting point — expect to tidy timing and note lengths. Drums aren't transcribed yet. Dynamics are approximate. Works best on clean recordings. **Is there a better engine for piano or vocals?** Yes, on Creator and above: the piano and vocal specialists transcribe the piano and vocals stem. Converting a stem you already split costs 990 credits per minute of the stem; starting from a full song with a specialist selected costs 1,728 credits per minute, the split included. Every other instrument uses the general engine. **Is audio to MIDI free?** It runs on every plan from your credits. The free Basic plan's monthly credits cover about 5.2 minutes of full-song conversion. **Can an AI agent or my code convert audio to MIDI?** Yes. Call POST /api/v1/midi/convert with a file or URL and poll /api/v1/midi/status/{job_id}, or use the audio_to_midi tool on the AudioPod MCP server. ## Related - [Bass to MIDI](https://audiopod.ai/audio-to-midi/bass) — The most reliable instrument, with its own tuning. - [Guitar to MIDI](https://audiopod.ai/audio-to-midi/guitar) — Experimental: what works and what does not. - [Piano to MIDI](https://audiopod.ai/tools/piano-to-midi) — The piano specialist engine (Creator+). - [Vocals to MIDI](https://audiopod.ai/tools/vocals-to-midi) — The vocal melody specialist (Creator+). - [Stem Splitter](https://audiopod.ai/features/stem-splitter) — Split first, then convert any pitched stem. ## Pricing Plan prices: [pricing](https://audiopod.ai/pricing). Start: [Convert audio to MIDI](https://audiopod.ai/auth/signup?callbackUrl=%2Fdashboard%2Ftools%2Faudio-to-midi).