Engram sampler runs AI offline to glitch audio — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Thoughtful Things launched a Kickstarter campaign for Engram, an AI-powered sampler and groovebox that mangles incoming audio and hallucinates new sounds — explicitly framed as a tool for experimental, uncanny sound design rather than radio-ready tracks.
- Engram runs a custom-trained "tiny AI" model entirely on-device with no internet connection, making it a standalone hardware instrument rather than a cloud service.
- Thoughtful Things trained its audio models exclusively on open datasets containing only CC-BY or similarly commercially licensed audio and pledged to "never train" on pirated or stolen data.
- Founder Evan King demonstrated the device by voice-requesting "piano," which produced a glitchy, vaguely piano-like output he described as a "field recorder for latent space."
- Engram's firmware will be open-sourced, letting users tweak the device or load their own models — a design choice the company ties to circuit-bending tradition.
- Kickstarter pricing starts at $675 for a limited early run, framed as a 30% discount, with expected retail pricing of $850 to $900.
Why it matters: Engram's offline-only architecture and open-firmware stance position it as a deliberate counter to cloud-based AI music generators like Suno, which face ongoing lawsuits over training data. By training only on CC-BY licensed audio and shipping a hackable $675 box, Thoughtful Things is targeting experimental musicians wary of the "AI music" label.
Ask SkimNews
