Engram: An AI-powered experimental music sampler
A Kickstarter project turns AI hallucinations into glitchy soundscapes
1AI-powered experimental instrument
Engram is a new sampler and groovebox from Thoughtful Things that uses AI to transform audio inputs into experimental, glitchy sounds. Unlike commercial AI music tools, it focuses on pushing the boundaries of AI audio models rather than creating polished tracks. The device is designed for sound designers and experimental musicians who want to explore the creative potential of AI hallucinations. It's positioned as a 'field recorder for latent space', capturing the unpredictable outputs of AI models.
2Local AI processing
Engram runs a custom-trained 'tiny AI' model locally without internet connection, ensuring privacy and offline functionality. The company emphasizes ethical training, using only audio datasets licensed for commercial use (CC-BY or similar). This approach contrasts with many AI music tools that have faced criticism for using copyrighted material. The local processing also allows for real-time manipulation of sounds without latency, making it suitable for live performances and studio experimentation.
3Circuit-bending inspiration
The device draws inspiration from circuit-bending, encouraging users to tweak and even 'break' its AI models for creative effects. This philosophy extends to the planned open firmware, which will let users load custom models. In demonstrations, vocal inputs like 'piano' produce distorted, glitchy approximations rather than clean reproductions. This intentional imperfection creates unique sonic possibilities that commercial AI music tools typically avoid.
4Kickstarter launch
Thoughtful Things launched Engram on Kickstarter with early bird units priced at $675 (30% off expected retail). The limited production run suggests a boutique product rather than mass-market appeal. While exact retail pricing isn't confirmed, estimates place it between $850-$900. This positions Engram as a premium tool for professional sound designers and experimental musicians rather than casual creators.
5Ethical AI training
The company explicitly states it only trains models on commercially licensed audio data, avoiding the controversies plaguing many AI music startups. This ethical stance includes a commitment against using pirated or non-commercial content. Such transparency is becoming increasingly important as legal challenges mount against AI companies using unlicensed training data, as seen in recent lawsuits against Suno and others.
6Creative sound exploration
Engram encourages users to explore the 'broken' outputs of AI models as a creative feature rather than a bug. The device treats AI hallucinations as desirable artifacts, similar to how vintage synthesizers embrace circuit-bent distortions. This philosophy challenges conventional approaches to AI music generation that prioritize clean, predictable outputs. It represents a niche but growing interest in experimental AI applications.
7Custom model potential
The planned open firmware will allow advanced users to modify Engram's behavior or load completely custom AI models. This flexibility could make the device appealing to academic researchers and AI developers alongside musicians. The approach mirrors the modular synth community, where users often modify and share custom patches. It also future-proofs the hardware as new audio AI techniques emerge.
8Niche market positioning
Engram's focus on experimental sound design and premium pricing targets a specific professional niche rather than the mass market. This contrasts with push-button AI music services aiming for mainstream appeal. The device fills a gap for artists seeking hands-on control over AI-generated sounds, particularly in live performance contexts where predictability is less important than unique timbres and textures.
Original article: The Verge (theverge.com) — Engram: An AI-powered experimental music sampler