---
title: Scores
emoji: 🎼
colorFrom: pink
colorTo: red
sdk: docker
sdk_version: 3.48.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: Analyze music scores from PDF files
app_port: 7860
tags: [mcp-server-track, agents-mcp-hackathon]
---

Demo video
--
https://youtu.be/_6Gr_ldh5lk

Participants
--

* jiba21
* ElasticExplorer

Audiveris MCP Server
--
In this space you can find the code which implements the Audiveris MCP Server which receives a
URL for a PDF music score and converts it to MusicXML format, which can be analyzed from LLL models.

If you want to use it from Cursor, as an example, you just need to add to Cursor MCP Tools the server:


{
  "mcpServers": {
    "audiveris": {
      "url": "https://agents-mcp-hackathon-audiveris.hf.space/gradio_api/mcp/sse"
    }
  }
}
```

and the you can ask Cursor to use it with a prompt like:

Given a PDF music score URL use MCP tool audiveris to convert it to a MusicXML file. Download the resulting MusicXML file, unzip the file and load the XML contents to be analyzed and complete a basic analysis. Use as URL https://www.mutopiaproject.org/ftp/BeethovenLv/WoO59/fur_Elise_WoO59/fur_Elise_WoO59-let.pdf

and you will get a basic analysis of the music score

```
Based on the MusicXML file analysis, I can provide you with a basic analysis of Beethoven's Für Elise:

Technical Details:
The piece was converted using Audiveris 5.6.1
It's written for Piano (Acoustic Grand Piano)
The MIDI channel is set to 1 with a volume of 78

Musical Structure:
Time Signature: 3/8 time
The piece begins with a pickup measure (anacrusis)
Initial dynamic marking: pp (pianissimo)

Opening Melody Analysis:
The piece begins with the famous E5-D♯5 motif
...
```