Show HN: Ableton Live MCP

123 points by bschoepke 67 days ago | 79 comments

Ever wanted to control Ableton with just your voice? Me too! I made this MCP server so I could just ask Codex to do anything in Ableton Live for me, while I was nap-trapped by my baby.

The chat messages I sent to Codex to make this:

in ableton, make a self reflective song, with audio vocals (via macos say) and chip tunes and 80's drum machines. should be a real edm banger

i want midi for everything but vocals please, with ableton devices. not prerendered audio for instruments

needs some fills

and should hit way harder after "3-2-1 i become the sound"

the vocals are squished too much (read too quickly), give them a little more length

add some dynamics, the song is basically one volume. and some pumping side chain

improve dynamics of the clap, seems a bit flat and indistinguished, want it harder after the 3-2-1 drop

introduce a new element on a new track after the 3-2-1 drop, that comes in but then recedes before the final exit

doesn't seem like the new thing has any notes

the element is a bit muddy/indistinct. perhaps it needs simplification and more space, different instrument choice, i dunno

ssalka 67 days ago |

Things I would use AI for in music production:

1. Generating track layouts (add tracks + empty audio/midi clips throughout)

2. Generating MIDI sequences

3. Generating Serum patches

4. Extracting stems from existing audio

5. Automating common workflows (eg sidechaining)

6. Semantic search of sample library

That being said, I don't think I want a full agentic workflow for vibe-producing. Point solutions seems like a better fit for me, personally.

bschoepke 66 days ago | |

> 6. Semantic search of sample library

I just pushed another tool for that! It wasn't exposed in the Live API but the implementation just issues the same queries to the underlying sqlite db that the Live GUI queries for the "Find Similar Sounds" feature.

dmje 67 days ago | |

I’ve had some fun building simple instruments in the browser using AI and piping midi to Live, then munging from there [0]. The whole principle of fully AI generated music leaves me cold but AI as a sort of sidechain to the creative process seems potentially interesting.

[0] https://variousbits.net/2026/02/22/building-generative-music...

bschoepke 66 days ago | |

Same here! I tried all of that, have 1, 2, and 5 working. So far it doesn't seem like Ableton's stem splitter or semantic search are programmatically accessible, but I didn't try very hard. I do have Serum so maybe I'll look into its file format; that does seem doable. The MCP already enables the agent to make patches for built-in Ableton devices.

corysama 67 days ago | |

Generating MIDI sequences

https://www.muse.art/home

bbrwx 66 days ago |

Interesting work and I’m really interested to see where this direction goes.

One question I keep coming back to is where tools like this MCP go beyond templating, like what I already use heavily in Ableton Live. In music production, many tasks are repetitive but not identical and that’s exactly where something like this can shine.

At the same time, music is widely seen as a "manual" craft. Every step in the chain from playing an instrument to a final music piece has both a technical and a creative side, and part of the process is staying curious and critical about what could be improved / done differently in the next project. That makes it an open question where automation actually adds value versus where it takes something away.

Where I’d personally love to see AI make a difference is in audio engineering / post processing, which also requires a lot of creativity beside a solid fundament of experience to really excel. There’s often a big gap between a great musical idea and a polished mix or master. If AI could help close that gap and contextually help to improve on things like tone, space, EQ, and loudness would be hugely valuable.

But the key for me is trust and transparency about modifications. I don’t want a black box that just makes things "better™." I’d want something that clearly explains its actions, like: "I've added an EQ to the piano on track 3 at 01:23 to open it up for the bridge so it sits better in the mix."

That kind of assistive, explainable approach would feel much more aligned with how people actually would be open for an assistant to create music.

bschoepke 66 days ago | |

Thank you, this gave me an interesting idea: while it can already set up a mastering chain for you, it is missing the full feedback loop of actually analyzing the processed audio signal. I think I'll add a small Max4Live device that enables the agent to probe the signal chain wherever it wants; from there it can write python to analyze the audio.

And yea, I kept the MCP general purpose so you can use it however you want. You can use it to simply ask questions about your live set; it can see the arrangement, see all the settings of all your devices, see all the MIDI, etc. I had it "fix a discordant sounding section" of one of my songs; it was like "oh yea tracks 3, 5, 6 all had some unfortunate notes colliding"

bschoepke 66 days ago | |

Added an "Agent Audio Tap" Max for Live device that lets the agent analyze any part of the signal chain and optimize your mix based on that (or just make cool visualizations!)

https://github.com/bschoepke/ableton-live-mcp/blob/main/READ...

breakall 67 days ago |

Yes! I want this for MainStage -- this would allow me to automate my weekly template setup for playing at my church. Each week before practicing I look up the songs in Planning Center and create a new MainStage concert file with one patch per song, and add notes to each patch screen with the song's key, etc. Automating this would save me the time of doing the busy work and get right to practicing.

bschoepke 66 days ago | |

Interesting use case! I'm not familiar with MainStage, but that workflow is possible with Ableton using this MCP. Codex itself can pull info from websites or other apps using its existing toolset and then push it into Ableton using this MCP.

Footprint0521 66 days ago | |

We should talk! I have a cli that does exactly this, I was able to reverse engineer the MainStage concert format

wartywhoa23 67 days ago |

> Ever wanted to control Ableton with just your voice?

Never.

jrajav 67 days ago | |

I guess the guidelines don't apply to you, as long as you disagree vehemently enough with the OP's basic intent.

wartywhoa23 67 days ago | | |

No amount of guidelines will make me lie in my replies.

windowliker 67 days ago |

For me, the point of making music is making it myself. If want to have something done for me I could just play someone else's record and pretend like I made it.

deng 67 days ago |

> should be a real edm banger

I'm afraid Codex ignored that one.

jhurliman 67 days ago |

Very cool! I posted my own experiments in this area a few months back, which were an iteration on an existing Ableton MCP. It’s great to see more people experimenting in the spaces of interfacing with complex applications and music production.

https://news.ycombinator.com/item?id=46428922

bschoepke 66 days ago | |

Nice, the audio analysis M4L device is something I wanted to add next, as that would close the feedback loop and let the agent actually analyze its setup. Kind of like how giving agents screenshots of its UI work helps improve the UI.

xrd 67 days ago |

Does anyone know of other MCP servers for similar music creative tools? I'm interested in things like sonic-pi, strudel.cc and orcas. But very open to anything. I think there is a good opportunity for kids to learn using these tools, especially if I can wire it into my mycroft.ai/neon device.

PaulDavisThe1st 67 days ago |

MCP for Ardour was added more than a month ago, thanks to contributor zabooma:

https://github.com/Ardour/ardour/commit/d582a0b042a68ccb22c0...

tim-projects 66 days ago |

I wish there was something like this for flstudio.

I've got 25 years of loops that basically to finish them need better arrangements. Using AI to auto generate sections is what I'm missing.

deng 66 days ago | |

> I've got 25 years of loops that basically to finish them need better arrangements

Welcome to the club. You need to learn how to actually finish a track, which is the most difficult but also the most rewarding part. Why would you use AI for that? I mean, just listen to that demo track Codex made in the above repo, you surely don't want that.

There's a good book about this, published by Ableton, you can read it for free here:

https://cdn-resources.ableton.com/resources/uploads/makingmu...

bschoepke 66 days ago | | |

To be fair, the demo track was one of the first I had it make, and I didn't put much effort into it because I thought it was especially funny with the macos "say" command vocals.

It's a garbage-in, garbage-out situation. If you give it more musical direction you will get more out of it.

2bewithu 66 days ago | | |

on a side note, his early 2000s output is genius

tim-projects 66 days ago | | |

> Welcome to the club. You need to learn how to actually finish a track

I don't think you understand. I've got thousands of songs. Why would I use Ai to generate arrangements... Maybe for ideas?

Maybe because certain things I'm lazy about?

Maybe because I've got thousands of songs?

It's not actually difficult to finish a song if your output is high enough. Sometimes the songs just come out without any struggle. But most of the time they don't.

I wrote and finished my first song around 1996. Using Cakewalk plugged into a midi keyboard.

HN is full of people who think using AI means you are lazy or can't do something. The fault is yours not mine. Adapt

olelele 66 days ago | |

Consider taking lessons from a human?

bel8 66 days ago |

This is really cool!

This kind of automation will allow impaired people to have access to a whole new world of creation. Blind and motor impaired comes to mind.

ktbwrestler 67 days ago |

this is awesome. Does anyone recommend one for Logic Pro X? I see a few in the wild but would love to help support if anyone is tinkering with one

rlupi 66 days ago |

Very nice! Thanks for sharing.

netics01 66 days ago |

Thank you for sharing!

markalby 67 days ago |

is this using M4L or the LOM ?

fassssst 67 days ago | |

Object model

maxbendick 66 days ago |

Careful you don't fuck up your hearing with an fx chain mistake.

robotswantdata 67 days ago |

If you’ve gone to the trouble of setting up Ableton MCP, you’ve already worked harder than Suno requires to make a banger

bschoepke 66 days ago | |

Well, it's also interesting to take a rough sketch out of Ableton, render it to audio, and then use Suno to "upscale" it.

gavmor 66 days ago | |

Suno is fun, but has absolutely nowhere near the affordances most musicians want.

Being a banger is not enough.

justinclift 66 days ago | |

Suno's output is grating though?

cyclopeanutopia 67 days ago | |

dyauspitr 67 days ago | |

Honestly, music generation is solved. You don’t have fine control at this point but people are unable to tell the difference anymore. There are tons of YouTube videos with blind tests between real artists and AI and people have no idea.

wartywhoa23 67 days ago | | |

The Matrix wants everything solved, doesn't it?

siquick 67 days ago | | |

Solved !== Good