Free online OPUS to MIDI converter

What the OPUS format is

OPUS is a lossy audio codec designed for speech and music, commonly stored in .opus files or inside Ogg and WebM containers. Web browsers, WebRTC applications, Discord, voice recorders, messaging applications, and ffmpeg-based tools commonly create or use it. OPUS files contain sampled sound; they do not contain the discrete note events that MIDI represents.

What the MIDI format is

MIDI is event data containing messages such as note number, velocity, channel, instrument program, and timing. A MIDI file contains no recorded sound, so its playback depends on the synthesizer, virtual instrument, or hardware receiving it. MIDI is useful for editing individual notes, changing instruments, importing melodies into notation software, and controlling a DAW.

How to convert OPUS to MIDI

Use an audio-to-MIDI transcription tool rather than a conventional file converter. The free Basic Pitch model is suitable for melodic and polyphonic music.

1. Install ffmpeg and Python, then install Basic Pitch with python -m pip install basic-pitch.
2. Decode the OPUS audio to WAV:

ffmpeg -i input.opus -ar 44100 -ac 1 input.wav

3. Run transcription:

basic-pitch midi_output input.wav

Basic Pitch writes a MIDI file and analysis files to midi_output; use the generated .mid file in a DAW or notation application. For a no-install attempt, you can upload the OPUS file to the free converter on 101convert.com, if its current tool supports audio transcription.

Quality and limitations

The result is an estimated transcription, not a recovery of original MIDI data. A clean, isolated melody produces better notes than speech, drums, reverb-heavy recordings, or a dense mix. Basic Pitch can detect several simultaneous pitches, but it may add false notes, merge notes, miss quiet notes, and misjudge timing or octave.

OPUS compression, background noise, clipping, and lossy speech encoding reduce transcription accuracy. Inspect the MIDI in a piano roll and correct pitch, note lengths, quantization, velocity, and unwanted notes manually. If the source is speech or ordinary environmental audio, the output will generally not be musically meaningful.