Guide · 4 min read

DreamSampler — How to Use

Output-only instrument that plays a live, cloud-generated buffer. Slice it up, play it chromatically, or fire one-shots — the engine keeps regenerating audio under your fingers.

DreamSampler is a VST3/AU instrument (and standalone app). It doesn't process your track's audio — it's output-only and plays a live, cloud-generated audio buffer that you slice up or play like a keyboard. The engine continuously regenerates that buffer, and it re-focuses generation under whatever you're currently playing.

Download DreamSampler

Setup & connect

1Add DreamSampler to an instrument track (or run the standalone app).
2Click Sign in. Your browser opens the Daydream sign-in page — sign in (it's free), then return to your DAW. The plugin signs itself in; there's no key to copy or paste.
3Load a source: drop an audio file on the Source dropzone, hit Upload, or pick one of the built-in sources when prompted (a basic beat or a lo-fi loop).
4Click Connect. The live buffer fills and its waveform appears. Everything below plays that buffer.

Shaping the buffer

Prompts: Prompt A and Prompt B describe what the engine generates; the Blend control morphs between them.
Strength, Structure, Timbre: Strength sets how hard the engine reworks the source; Structure how closely it follows it; Timbre how much of the source's character survives.
Styles and references: Add a Style to steer the aesthetic, and drop optional Timbre Ref / Semantic Ref clips to weight the Timbre and Structure knobs.

The three modes

Slice (default): Chops the buffer into pads mapped to consecutive keys from C1. Trigger from on-screen pads, your computer keyboard, or MIDI.
Classic: Chromatic keyboard — pitches the whole buffer up/down by note and sustains while held. MIDI-driven; no on-screen pads.
1-Shot: Each note plays the entire buffer once, start to finish. MIDI-driven; no on-screen pads.

Classic and 1-Shot have no on-screen pads — drive them with a MIDI clip or controller into the instrument track. Only Slice is playable directly from the plugin window.

Slice mode (the default). On first buffer fill it auto-slices so the kit is playable immediately. Slices follow the morphing buffer — the engine keeps regenerating audio under the pad you're holding.

Three ways to trigger a slice

On-screen pads: Click a pad on the waveform.
Computer keyboard: The z x c v b n m , and a s d f g h j k keys play the first 16 slices.
MIDI: Slices map to consecutive keys starting at C1 (MIDI note 36).

Slice By — how the buffer gets chopped

Transient: Auto-detect hits; use the Sensitivity knob to get more or fewer.
Beat: Slice on a musical Division — 1/4, 1/8, 1/16, or 1/32.
Region: A fixed number of equal slices, set by the Regions knob (1–64).
Manual: No auto-slicing. Double-click a slice to split it at the click point, and double-click a marker to merge two slices.

Classic mode. A chromatic instrument: each MIDI note plays the whole buffer transposed (relative to the root) and sustains while the key is held. Driven by DAW MIDI — a clip or your controller into the instrument track.

1-Shot mode. Each MIDI note plays the entire buffer once, start to finish — ignores note-off. Also MIDI-driven.

Shared sound controls

Gain and Volume: Input Gain into the voice and output Volume level.
Play and Playback: Play sets Trigger or Gate behaviour; Playback is Poly/Mono — locked to Mono for now (Poly is coming).
Fade In / Fade Out: Per-voice amplitude ramps. Fade Out begins before the end so the tail stays audible.
Transp: Global pitch offset across the kit, in semitones.
Vol < Vel: How much MIDI velocity scales level.
Filter: On/off, type (LP / HP / BP), slope (12 or 24 dB), Freq (cutoff), and Res (resonance).

Good to know

One live pad at a time: The cloud has a single generation playhead, so only one pad or voice is "truly live" at any moment — polyphony is planned.
Nothing sounds until the buffer fills: Sign in, load a source, and Connect first. The empty-waveform hint tells you what's missing if a pad isn't responding.
The waveform playhead doesn't sweep on its own: DreamSampler is a sampler, not a player — motion comes from triggered voices.