Audio to Band Score skill
A skill that turns a song recording into a five-piece school band arrangement: MIDI, PDF and an editable MuseScore file
- 237 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Scripts read your audio files and write separation and export results to disk, and the environment is set up by hand with external packages.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/audio-to-band-scoreDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Scripts read your audio files and write separation and export results to disk, and the environment is set up by hand with external packages.
This is third-party code. Review the repository files before installing.
What it does
The skill guides an agent through the full path from a mixed recording to band charts. It splits the audio into time-aligned stems with Demucs and identifies song sections, tempo, time signature, harmony, bass movement, riffs and entry points. It then distributes parts between two guitars, bass, keyboard and drums with playability in mind. The arrangement can be tuned to a level: beginner, school amateur intermediate or advanced. The result is built as a structured editable MuseScore file rather than raw recognition output. Scripts audit score structure, string and fret relationships in tablature, export safety and file consistency. The output is MIDI, a print-ready PDF and an `.mscz` file. The vocal track is used only to understand structure, breathing space and entry points and is not added to the score by default. Processing runs on local free tools and audio is not sent to a paid external API.
Who it is for. School and amateur musicians and band leaders who need a playable arrangement of a favorite song without transcribing parts by ear.
Good fit when
- You need a score for lead guitar, rhythm guitar, bass, keyboard and drums from a finished recording.
- Parts have to be simplified for the players level: beginner or school.
- You need a file to edit in MuseScore, not only a PDF.
- You want to split a recording into stems for analysis on its own: `scripts/separate_audio.py` fits.
Not a fit when
- You need a score app with a window: this is an agent skill and not a standalone program.
- You need an exact note-for-note transcription without edits: the release is called early and the result needs review.
- You have no MuseScore Studio and FFmpeg: export and decoding will not work without them.
Example request
Use `$audio-to-band-score`: arrange `song.wav` for lead guitar, rhythm guitar, bass, keyboard and drums and deliver MIDI, PDF and MSCZ.Limitations
The project is called an early release. The workflow does not install software or download model weights by itself: prepare an environment with Python 3.11 and `demucs`, plus FFmpeg and MuseScore Studio. The separation output directory must be new, existing directories are never overwritten. The result stays a draft to check by ear, especially in complex rhythms. Rights to the source recording and arrangement stay with the song rights holders, and compliance rests with the user. The agent is connected by making the repository available to its runtime; there is no separate install command.
How to disable. Delete the cloned repository folder from the agent skills directory and the `work/audio-env` virtual environment.
FAQ
How do I prepare the environment?
Create a Python 3.11 virtual environment and run `python -m pip install demucs numpy soundfile imageio-ffmpeg`, then add MuseScore Studio.
Which parts do I get by default?
Lead and rhythm guitar in tablature, bass in tablature, keyboard on a grand staff and drums in standard notation.
Do I need a GPU?
No. CUDA is optional and without a compatible GPU processing runs on the CPU.
How do I split a recording into stems on its own?
With `scripts/separate_audio.py` and the `--input`, `--out-dir`, `--profile high` and `--device auto` flags.
What is the difference between `high` and `balanced`?
`high` runs four-source and six-source separation, while `balanced` runs six-source only and suits trials and modest machines.
Discussion
No questions or reviews yet.
Related
SkillsOriginal instrumental music written as notes, with no audio generator: the agent composes, the skill reviews and mixes.
29 skills for composing and arranging pop music: from a brief the agent fills in a spec sheet for Suno, YuE2 or MIDI
A local studio plus skill: the agent writes the score and real instrument recordings play it, with no music-generation model.
An MCP server driving an open MuseScore score: the agent writes notes, lyrics and repeats and adds color markup.
MCP server for a Windows music studio: the agent writes lyrics, generates a song locally, arranges, mixes and exports WAV
A skill that makes a music video for your song: every frame is drawn by code and six steps with pauses lead from analysis to edit
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Audio to Band Score (this card) | no | 237 | ||
| MuseScore MCP Server | no | not required | 145 | MuseScore MCP Server drives MuseScore from an agent, while this skill produces an arrangement from a recording with audio separation. |
| music-composition-skills | no | 209 | music-composition-skills helps compose music, while here a finished song recording gets arranged. |
README in short
The README lists capabilities, default instrumentation, workflow steps and a quick start. MIT license, version 0.2.0 added a score event model and independent rhythm and fidelity audits.