Mirelo

Convert a musician's recording into editable MIDI and MusicXML. Default timing preserves the performance; set timing to quantized only when the musician asks to snap notes to the beat. Set score_pdfs for engraved sheet music and optional tablature. This is a paid asynchronous job: use preflight after inspect_asset, then wait_job or get_job. Pass idempotency_key when retrying the same request.

audio_to_midi

Inline upload, kept for small files only: the bytes have to be emitted into this conversation, so it is capped at ~96 KB decoded. Use create_upload for external media of any real size. The url form accepts only a Mirelo download_urls link from a prior job. The returned asset_id can go straight to a generate or edit tool; video_to_sfx measures the clip itself, so duration_ms can be omitted.

create_asset

Preferred way to send external media. Returns a pre-signed storage URL so the bytes go straight from disk to storage without passing through this conversation — run the returned upload command in a shell, then pass asset_id straight to a generate or edit tool: video_to_sfx measures the clip itself so you can omit duration_ms, and the extend tools probe the audio prefix for you. inspect_asset is optional and only worth a call when you want the metadata before committing credits. Download remote media to a local file before using this tool; hosted MCP accepts url only for a Mirelo download_urls link from a prior job. Two commands come back, one per shell family: run curl_command in a POSIX shell (bash, zsh, Git Bash, or pwsh on macOS/Linux) or windows_powershell_command in Windows PowerShell. They quote differently and the wrong one will not work, so pick by the shell you are about to run and paste it as given rather than rewriting it — it already sets the Content-Type the URL was signed for, and a PUT that sends a different one (or none) is rejected outright. Files up to max_size_bytes work on this path. The URL stops working after expires_in seconds, so upload right away and call create_upload again if it lapses. If the PUT fails with a proxy error, your shell has no outbound network — that is a client setting, not a Mirelo error, and retrying will not help. Either ask the user to allow outbound requests to the upload host and start a NEW conversation (the setting does not apply to the current one), or ask them to upload the file at https://mirelo.ai/studio/mcp-upload and paste back the asset_id it shows. Falling back to create_asset only works for files of a few KB.

create_upload

Submit text-conditioned audio extend as an async job (v1.6). Returns job_id; use wait_job (or get_job) until succeeded/errored. Requirements: input audio prefix must be ≥ 3s; append_duration_ms 1000–57000 (≥ 1000 when loop=true). Provide exactly one of append_duration_ms or duration_ms. You never need to measure the input audio: pass append_duration_ms (how much new audio to add) and the server probes the prefix with ffprobe and derives the total itself. duration_ms is the legacy total (prefix + extension) and is the only reason to know the input length — prefer append_duration_ms and do not shell out to ffprobe. Use asset_id from create_upload for external media. The url form is reserved for a Mirelo download_urls link from a prior job. Chaining: pass a prior job's download_urls entry straight back as url. Raw result_urls remain download fallbacks, but they are not accepted as hosted MCP inputs. Treat a download_urls entry as audio unless you know that specific artifact is a muxed video file. Requesting output:"video" does not guarantee one — a long clip returns audio anyway — so pass a result as a video url only when the result itself is video. Inline base64 on create_asset is a poor fit for audio: uncompressed WAV at the shortest length these tools accept is already well over the inline limit. A short, heavily compressed clip can fit, but use create_upload rather than shrinking audio to clear the limit, since the waveform is what the model conditions on. Pass idempotency_key on retries to avoid a second paid job.

extend_audio

Submit video-guided audio extend as an async job (v1.6). Returns job_id; use wait_job (or get_job) until succeeded/errored. Requirements: input audio prefix must be ≥ 3s (≥ 3000 ms); append_duration_ms 1000–57000. Provide exactly one of append_duration_ms or duration_ms. You never need to measure the input audio: pass append_duration_ms (how much new audio to add) and the server probes the prefix with ffprobe and derives the total itself. duration_ms is the legacy total (prefix + extension) and is the only reason to know the input length — prefer append_duration_ms and do not shell out to ffprobe. Use asset_id from create_upload for external media. The url form is reserved for a Mirelo download_urls link from a prior job. Chaining: pass a prior job's download_urls entry straight back as url. Raw result_urls remain download fallbacks, but they are not accepted as hosted MCP inputs. Treat a download_urls entry as audio unless you know that specific artifact is a muxed video file. Requesting output:"video" does not guarantee one — a long clip returns audio anyway — so pass a result as a video url only when the result itself is video. Inline base64 on create_asset is a poor fit for audio: uncompressed WAV at the shortest length these tools accept is already well over the inline limit. A short, heavily compressed clip can fit, but use create_upload rather than shrinking audio to clear the limit, since the waveform is what the model conditions on. Pass idempotency_key on retries to avoid a second paid job.

extend_audio_with_video

Get the authenticated Mirelo account’s available credits, shortfall for a minimum billable request, and provisioning state/deadline.

get_account

Poll status for a job_id from text_to_sfx / video_to_sfx / extend_audio / extend_audio_with_video / inpaint_audio / audio_to_midi. Prefer this when the client can drive its own poll loop. On success, prefer `result.download_urls` (short links on mirelo.ai) over `result.result_urls`: the raw URLs are ~500 characters, which some clients reject, and they expire an hour after generation regardless of when you read them. A download link is signed when followed, works for the whole session, and can be passed straight back as an input url to extend_audio / inpaint_audio. `download_urls` is index-aligned with `result_urls` when present. Raw `result_urls` remain output/download fallbacks, but hosted MCP does not accept them as input URLs; if a download_urls link is missing, download the result and stage it with create_upload instead.

get_job

Read completed Audio-to-MIDI note data in pages. Use this only when note-level data is needed for MIDI visualizations or editing; job results intentionally return a compact summary instead.

get_transcription_notes

Submit audio inpaint as an async job (v1.6). Optional video guides the replacement. Returns job_id; use wait_job (or get_job) until succeeded/errored. Requirements: start_ms ≥ 1000; gap width (end_ms − start_ms) between 1000 and 8000 ms. Use asset_id from create_upload for external media. The url form is reserved for a Mirelo download_urls link from a prior job. Chaining: pass a prior job's download_urls entry straight back as url. Raw result_urls remain download fallbacks, but they are not accepted as hosted MCP inputs. Treat a download_urls entry as audio unless you know that specific artifact is a muxed video file. Requesting output:"video" does not guarantee one — a long clip returns audio anyway — so pass a result as a video url only when the result itself is video. Inline base64 on create_asset is a poor fit for audio: uncompressed WAV at the shortest length these tools accept is already well over the inline limit. A short, heavily compressed clip can fit, but use create_upload rather than shrinking audio to clear the limit, since the waveform is what the model conditions on. Pass idempotency_key on retries to avoid a second paid job.

inpaint_audio

Read what an uploaded asset actually contains: duration_ms, content_type, size_bytes, plus width and height for video. Optional before generating — video_to_sfx measures an asset_id itself, and the extend tools probe the audio prefix — so call it when you want to confirm the create_upload PUT landed, or to see the duration and dimensions before committing credits. Each call spends one ffprobe pass, so do not call it as a routine step before every job. A not-found error means the PUT has not completed. A null duration_ms means we could not read one: usually the file is not decodable media, so re-check it before retrying, and supply duration_ms yourself if you know it.

inspect_asset