MIDI MIME Type

What does audio/midi mean?

The audio/midi media type is commonly used for Standard MIDI Files, or SMFs. These compact binary files store time-stamped performance instructions such as note events, channels, tempo changes, controller movements, track names, and lyrics. They do not store a recorded waveform.¹

That distinction explains why the same MIDI file can sound different on two systems. A player must send its events to a synthesizer, sampler, sound module, or virtual instrument. The chosen instruments, sound bank, effects, and synthesis engine determine the resulting sound.²

How to identify a Standard MIDI File

A normal SMF starts with a header chunk whose first four bytes spell MThd:

Track chunks follow with the identifier MTrk. Each holds delta times and MIDI or meta events. The MThd signature is more reliable than the extension alone, although a complete validator must also check chunk lengths and event encoding.⁶

The familiar extensions are .mid and .midi. Some applications also use .smf, but extension support is inconsistent. Ableton Live, for example, imports .mid and .smf yet tells users to rename .midi files to .mid before importing them.⁷

Where Standard MIDI Files came from

MIDI 1.0 began as a hardware and message protocol for exchanging performance data between instruments. The first public connection demonstration took place in January 1983, and an initial MIDI 1.0 specification was available that August.⁸

Sequencer programs still saved projects in incompatible proprietary formats. The MIDI Manufacturers Association adopted the Standard MIDI File format in 1988 to make time-stamped MIDI performances portable between programs and computers. Its design favors compact interchange rather than replacing every application's richer project format.⁶⁹

SMF remains useful for exchanging editable notes, tempo maps, controller automation, karaoke lyrics, and arrangements. It is not a complete production session. Audio recordings, plug-in settings, proprietary instrument patches, mixer routing, and many application-specific editing details usually stay behind in the original project.

What are MIDI formats 0, 1, and 2?

All three SMF variants use the same MThd and MTrk chunk structure. The format value in the header changes how the tracks relate to one another:⁷

Converting format 1 to format 0 can combine tracks while preserving channel events, but the original track organization may be lost. Converting the result back cannot reconstruct information that was discarded.⁷

General MIDI is a separate compatibility layer. It assigns conventional instruments and percussion to program and channel numbers so a suitable synthesizer can make more predictable choices. An SMF need not follow General MIDI, and Ableton Live does not automatically assign sounds from General MIDI program data.⁷

How to open an audio/midi file

Use a sequencer, notation program, DAW, or dedicated MIDI player with a synthesizer:

• Apple Logic Pro can open, import, and export SMF format 0 and format 1 files. When opening one, it can create software-instrument tracks for the MIDI tracks.¹⁰
• Ableton Live imports formats 0, 1, and 2 and exports format 0. A MIDI track still needs an instrument before it produces sound.⁷
• Apple's AVMIDIPlayer API plays MIDI data through a system sound module and can use a supplied sound bank.¹¹

Current browser media elements should not be assumed to play SMF directly. Chromium's native playback list includes formats such as MP3, FLAC, Opus, Vorbis and Ogg, PCM, MP4, WebM, and WAV, but not MIDI synthesis.¹² The Web MIDI API is different: it provides access to connected MIDI input and output ports, not a built-in Standard MIDI File player or sound bank.¹³

Support in Poly

Poly defines an explicit audio/midi type and its client treats it as audio. Filename inference is narrower than the common MIME mapping: Poly recognizes .midi, but not .mid, even though both extensions are widely used.

Recognized MIDI files enter Poly's generic audio path. That path expects a decodable audio stream and uses Symphonia to probe containers, extract duration and tags, prepare audio-similarity segments, and enable transcription. Symphonia has no Standard MIDI File demuxer or synthesizer, so an SMF cannot complete that processing path. The file is still stored, synced, and searchable by filename and ordinary properties.

Poly also routes recognized .midi files to its in-app audio player. The player supplies the original file URL to the browser's <audio> element. Because the file contains events rather than PCM or encoded sound and current browsers do not provide native SMF synthesis, playback normally fails. Poly does not currently render the MIDI through a sound bank or generate a browser-friendly audio copy.

The Poly agent cannot read the musical event stream. There is no transcript when audio decoding fails, and the agent's direct-audio allowlist covers WAV, MP3, AIFF, AAC, Ogg, and FLAC rather than audio/midi.

Converting MIDI to audio

Converting MIDI to WAV, FLAC, MP3, AAC, or Opus means rendering it through a synthesizer. The output depends on the chosen sound bank, instruments, tempo interpretation, controller support, and effects. It is not a format-only transcode.

Choose the result based on the next step:

• WAV or FLAC for a lossless capture of a chosen rendering.
• AAC, MP3, or Opus for smaller playback copies.
• The original SMF when notes, tempo, channels, or controller events must remain editable.

Listen to the entire render before discarding anything. Missing sound banks, unsupported System Exclusive messages, different General MIDI implementations, or ignored controller data can change the arrangement substantially.

The reverse operation is also different. Converting recorded audio to MIDI requires note and rhythm transcription. It is an analysis task, not a lossless recovery of the original performance instructions.

Is audio/midi the same as audio/midi-clip?

No. IANA registered audio/midi-clip in 2024 for the MIDI 2.0 Clip File format. It stores Universal MIDI Packet data, uses the .midi2 extension, and begins with the eight-byte signature SMF2CLIP. The registration explicitly distinguishes it from the older SMF formats 0, 1, and 2.¹⁴

Likewise, audio/rtp-midi identifies real-time MIDI commands carried over RTP, not a downloadable SMF. A matching word in the media type does not make these formats interchangeable.

Security and privacy considerations

Standard MIDI Files are data, not executable programs, but parsers and synthesizers still process untrusted binary input. Validate chunk lengths, cap event counts and playback duration, and use maintained software. Malformed delta times or dense event streams can consume excessive CPU or memory in an incautious implementation.

SMFs can also contain text, track names, lyrics, copyright notices, and System Exclusive events.¹⁶ Review those fields before publishing a file. Rendering to audio may hide the embedded text from casual inspection while leaving it intact in the original.