One of the latest projects on Kickstarter is Engram, a generative audio sampler and groovebox, developed by Thoughtful Designs.
The campaign claims this is a new kind of music device, combining a traditional sampler workflow with hands-on sample manipulation using tweakable tiny audio models.
The philosophy is to use AI and machine learning in a human-tweakable way, rather than being hidden as effective black boxes which churn out results but have very little creative input from real people.
Lead developer Evan King suggests that tools need to be explorable, tweakable, and breakable. He points to the past, where equipment such as amps were overdriven and turntables were effectively abused to create new sounds.
AI still makes a lot of people cringe or turn off altogether. This project claims it only uses models trained on open datasets licensed for commercial use. Hand-designed tiny AI models generates sound from custom-trained audio models.
Key features of the Engram include:
- Neural audio processing – process and warp your own recordings with an embedded neural audio codec
- Sample generation – create and glitch short bursts of audio to sample and manipulate
- Model bending – unique, tweakable features for glitchy neural audio processing
- Plug-in audio models – create, share, and plug-in new audio models to experiment with, just like old-school ROMpler cartridges
- Sample and pattern storage and recall
- MIDI input and output
- CV sync/gate input and output
- Stereo line-level input for sampling
- Headphone output
- On-board microphone for lo-fi sampling and voice control
A prototype has already been built and is in testing.
