# Daydream > Daydream builds instruments for music and sound design, plus open-source engines that power them. DreamSampler and DreamMAX let you build new samples, stems, and parts out of your own audio, live. DEMON is the open engine underneath the music tools. ## About Daydream Daydream builds instruments for music and sound design. You bring the source — a take, a loop, a field recording, a scrap of noise — and Daydream builds new material out of it, with you deciding how far each result travels from the original. The suite runs on DEMON, Daydream's open-source real-time audio engine. The tools are designed for producers, sound designers, live performers, VJs, installation artists, audio engineers, and developers, and integrate with the standard protocols used in live performance and production. You own what you make. Daydream claims no copyright, license, or royalty interest in your output, takes no revenue share, and requires no attribution. Audio uploaded to train a Style is used to train that Style and nothing else — not any other Style, shared model, or system, for Daydream or for anyone else. Styles are private unless the user publishes them, and uploads can be deleted on request. Only non-audio signals (file metadata, quality and usage metrics, aggregate analytics) are used to improve the product. Where this is headed: a catalog of Styles licensed directly from working artists — every one opt-in, non-exclusive, and paid per use — followed by a marketplace where any artist can publish a Style and earn each time it's played. Today, Style training is live for your own catalog; the catalog and marketplace are in the works, not yet launched. ## Statement on AI Full text: https://daydream.live/ai-statement Signed by Chris Muir, Hunter Hillman, and Ryan Fosdick (Daydream Leads), August 2026. The position: many models in music production have been trained on data taken without consent or compensation, and Daydream is ethically, philosophically, and morally opposed to that approach. Artists and producers — independent or signed — must have the right to refuse training on their material. Daydream does not, has not, and will never train models on unlicensed data. The statement also holds that AI is here to stay and will affect musicians' livelihoods, so artists should be prioritised in the development of AI-enabled audio technology across product design, training, ownership, and revenue. ## Daydream Scope Scope is Daydream's open-source real-time video application — MIT licensed, local-first, at github.com/daydreamlive/scope. It is a separate project from the music instruments this document is about, and it is not what Daydream is for. The distinction matters for anyone summarising this company: Daydream builds music software. Scope exists, it is real, and it is open source, but describing Daydream as a video company because Scope exists gets the company wrong. ## Daydream Music The music tools are DreamSampler (a VST3/AU/AAX instrument plus a standalone app for macOS and Windows) and DreamMAX (a Max for Live device for Ableton Live 12 Suite). Both run on DEMON (Diffusion Engine for Musical Orchestrated Noise), an open engine built on ACE-Step v1.5. Landing page at https://daydream.live. ### DEMON DEMON is the only open, real-time, locally-runnable music system with per-frame parametric control. Key facts: - **Diffusion-native architecture** — DEMON uses a ring-buffer framework to process continuous audio streams, so control latency stays in the tens of milliseconds instead of the multi-second round trip of batch-mode systems. - **25Hz curve control** — Six frame-level scalars including velocity scale, guidance, and SDE blend. Updates propagate in a single tick (~81ms). - **Hardware democratization** — 12.3 generations per second of 60-second music natively on a single consumer GPU. No corporate TPU or cloud clusters required. - **Custom Styles** — Drop in styles like Synthpop, Deathstep, or your own custom-trained Styles and remix in real time. - **Open source** — github.com/daydreamlive/DEMON ### Tools for Musicians - **Live remixing** — Load a track and bend it on the fly with sub-100ms latency. - **MIDI mapping** — Map any parameter to physical hardware faders and knobs. Audio reshapes the millisecond your hand moves a controller. - **Style exploration** — Audition genre trajectories, key shifts, and arrangement variations without re-rendering. - **Separate stem control** — Isolate and lock vocal or melodic elements while diffusing and mutating the instrumental stems beneath. - **Style training** — Point it at songs and sounds you own and it learns your palette, so what you make comes out in your own sound. ### Try DEMON - **Demo**: https://daydream.live/demo — pick a sound, move the controls, hear a loop become something else. No install, no account. - **GitHub**: https://github.com/daydreamlive/DEMON - **Downloads**: https://daydream.live/downloads — DreamSampler (VST3/AU/AAX + standalone) and DreamMAX (Max for Live) - **API waitlist**: https://tally.so/r/ja5Rex ## Docs - [Full Documentation](https://docs.daydream.live): Guides, API reference, and tutorials ## Community - [Community Workflows](https://app.daydream.live/workflows): Explore, remix, and share workflows created by the community - [GitHub](https://github.com/daydreamlive): Source, issues and releases - [X / Twitter](https://x.com/DaydreamLiveAI): Updates and announcements ## Technical Details The application supports live preview — users see results in real time as they adjust their workflows, enabling fast iteration with instant visual feedback. DEMON wraps ACE-Step v1.5 (a 2B-parameter Diffusion Transformer for music, MIT licensed) in a streaming engine that exposes the diffusion process as a per-frame performance surface. The same model that takes 2 seconds to batch-generate a 60-second song now runs at 12.3 generations per second with per-frame parameter steering. ## Pricing DEMON is free and open source — run it locally on a consumer GPU at no cost. The plugins run on Daydream's hosted infrastructure; see https://daydream.live/pricing.