Cochlea MCP server

Cochlea is a tool for making and studying audio without ever pressing play. You describe a piece of music as plain text, and Cochlea turns it into a real audio file, then measures what came out. It is handy if you work with sound, jingles, or audio checks and want your AI to handle the boring parts.

What this MCP server does

You ask your AI for something audio related, like rendering a short tune or checking how loud a file is. The AI sends that request to Cochlea through this helper. Cochlea writes the audio file, or opens the file you pointed at and studies it. Then it sends back numbers, a short text report, or a picture of the sound. Your AI reads that and tells you what it found.

Click to zoom

What you can do with it

• Render a music score you describe as text into a WAV file
• Measure loudness, pitch, tempo, and key of any audio file
• Draw a spectrogram picture of a sound so you can see it
• Compare two audio files and show what changed between them
• Import a MIDI file and turn it into an editable score
• Turn a recorded solo into a score file
• Check a score for mistakes before you render it

Try asking your AI

• “Render this score to mix.wav and tell me the loudness”
• “Probe my recording.wav and give me the tempo and key”
• “Make a spectrogram of bars 17 to 19 of this file”
• “Compare old_mix.wav and new_mix.wav and tell me what changed”

What it gives back to you

You get back short reports with numbers like loudness, tempo, key, and note names. For a spectrogram, you get a small picture right in the chat. For a render, you get a file saved on your computer plus a confirmation. Everything is written in plain text you can read.

Before you start

Who it's for

People who work with audio, like sound designers, music makers, podcast editors, or anyone who needs to check or create audio files without listening to them.