Do you want to use AI as a tool for synthesis or composition?
That’s the question raised by the latest sonicLAB update to Fundamental synthesizer. Fundamental 4.1 an update that lets you use AI prompts to accelerate your sound creation and composition.
Fundamental 4.1 now exposes its sound engine through the Model Context Protocol, so any MCP-compatible LLM client can drive synthesis directly. You can use it with tools like Claude Desktop and Gemini CLI.
Fundamental lets you enter in prompts, which are translated into presets using AI. Here’s an example of the type of prompt that they suggest you can experiment with:
“Set the oscillator frequencies based on the distance of the planets from the sun in our solar system. Find the other planet frequencies proportional to Mercury which is assigned to lowest frequency as 150hz and becomes the first oscillator. Change the gain of the oscillators continuously according to orbits of each planet around the sun. Make the Earth full orbit 8sec. long and assign the other planet orbits proportional to it.”
Pricing and Availability:
Fundamental 4 is available for Mac & Windows for $136.00 USD. The update is free for existing users.
No. Why should I? If I fail to come up with my own ideas I simply stop doing music. It’s quite easy.
Again with the AI posts? Read the room, synthhead.
No. These AI posts may mean I unsubscribe from Synthtopia, just like when I bulk unsubscribed from people peddling crypto and NFTs. Planet destroying snake oil BS.
For the most part, using AI for composition feels like wearing three condoms at once. I can see some interesting housekeeping uses for it, but I feel that I should turn the knobs until I find what I need. You need to be selective about where you use it. Making music shouldn’t be too easy.
Some people might argue that if turning knobs is involved, you’re not making music to begin with.
Nope. Getting my harmonies commented on is ok but I am in charge of my own glitches.
I don’t see why the hate. The video looks super cool. Great way to test some complicated intellectual ideas.
Just watched the video. Why in god’s name would I ever want to use such unmusical “sound effects” in my compositions? Is this supposed to be “Gas Music From Jupiter”? Plus, the second example effect has some really awful clipping.
I do not care to use an LLM to generate patches, but it’s worth noting that the Model Context Protocol is basically a JSON API to the synthesizer, meaning that if you are so inclined you can write your own scripts/software to control the synthesis engine. Normally to do this you would need to write code that generates MIDI and then point it at the plugin or device of your choice, but that will only as good as the synthesizer’s MIDI implementation (whcih as we know is often lacking).
The MCP means you can directly control all the internal parameters and they will be correctly names as oscillator_1.pulse_width.modulation_source or whatever. If you want to do super-complex sequencing/modulation but you don’t want to get into the weeds of doing fourier transforms and writing the DSP to synthesize the audio in real time then this could be very useful. You don’t HAVE to connect it to an LLM to do this, the point of the Model Context Protocol is that it’s open to the outside world.
HAving said that I think this trend is probably going to be become ubiquitous and there will be a lot of craptacular AI-prompted patch libraries generated by people who have no real interest in or experience at synth programming.