ChatGPT has been making headlines in every corner of the technological landscape this year, and the music tech world has been no exception. Since it launched, we've seen the AI-powered chatbot construct chord sequences, write lyrics, make playlists, and even produce code for effects plugins.
YouTuber Doctor Mix has added to this list of achievements by creating the first VST synthesizer from code authored by artificial intelligence. In a video posted yesterday, February 28, the YouTuber used ChatGPT to develop a synth plugin that he's (rather unimaginatively) named Doctor Mix AI Synth, and released to the public as a free download.
Though ChatGPT generated code used in the plugin, the process wasn't quite as simple as it sounds: after first asking the chatbot if it could code a synth, Doctor Mix received a negative response: "I'm sorry, but as a language model, I do not have the capability to code a VST synthesizer", ChatGPT responded.
With some cajoling, however, the chatbot was able to generate code for a simple sine wave synth to be deployed within the open-source C++ framework WDL-OL. Doctor Mix then attempted to use ChatGPT to guide himself through the process of transforming this code into a useable plugin, but shortly hit a brick wall.
After heading to Amsterdam to seek some assistance from plugin developers Martinic - and getting distracted by a few choice bits of vintage gear in the process - Doctor Mix was able to integrate the code generated by ChatGPT with some additional code written by Martinic's developers.
This process produced a working software synthesizer capable of receiving MIDI input. Martinic's team then added some extra features with the help of ChatGPT, including a filter and an amplitude envelope, compiled this code into a VST/AU/CLAP plugin (complete with a somewhat questionable GUI) and released it for free via their website.
While it's far from the most powerful free VST synth plugin we've ever come across, it's certainly the first to be co-developed by AI, and for that it deserves some recognition.