About

The band's rig.

JAMWire Studio is built for musicians by musicians, for DIY bands that run their own sound, and write their own songs. It exists because recording a real show (every input, every camera, in sync, without a crew) was still a pile of separate tools and a night of fixing it afterward.

What it's for

Point it at the room

Completely replace your old A/V drag and drop, chop and nudge workflow.
With JAMWire Studio, capturing the performance has never been easier. Every device captured together, on one clock, with the setlist, the charts and the lighting cues playing along in time. You take home a finished, perfectly synced, multimedia multitrack session of the performance that you just live streamed from the venue.

That's the job JAMWire Studio is built for. The DAW and NLE video editor are full-featured so that you never have to leave the workstation, and sessions are fully portable in case you still want to.

On the Network, only the Host Studio needs studio hardware. Everyone else in the band joins from a phone, a browser, or the DAW they already use. The pricing page explains who pays for what.

What it will never do

The short list

No dongles, activation servers or phoning home. The app works on your machine whether or not ours is reachable. Licensing is an honor system, the way REAPER does it.
Your work is never stuck here. Open formats throughout. Sessions export to REAPER, Final Cut Pro and OpenTimelineIO today, and Project Rosetta will make the converter open source.
The app only interrupts you about your work. Disk nearly full is a real warning. A subscription reminder is a dismissible splash you'll see once.
Nothing you record is uploaded unless you turn cloud storage on. And when you do, deleting anything is your decision, on a published schedule.
No ads or crippleware. The beta is the full app, and so is the evaluation after it.

How it's built

Low latency first, then everything else

The engine runs a fixed 32-sample block and is written to that ceiling. That limit decides how plugins are hosted and processed, how video frames move, how the timeline keeps time, and how two machines agree on the transport state with microsecond precision.

Everything shares one clock. Audio, video, MIDI, lighting and the timeline itself run on a timebase that divides cleanly to every common sample and frame rate; edits and punch-ins land on the same time scale no matter what type of media.

It's written in Rust, with the platform pieces (Core Audio, Metal, AVFoundation) behind a small native layer. Analysis features (stem separation, tempo detection, transcription, amp modeling) run in-process on your machine. Zero cloud-dependent processing.

What you'd expect from a DAW

Recording: Multitrack capture, punch in and out, pre-roll and count-in, loop recording into stacked takes, and comping across them
Editing: Razor, trim, join, fades, reverse, normalize, consolidate, keyboard nudge and ripple edits, with snapping to clip edges, markers, the playhead and frame boundaries
MIDI: Piano roll editing, quantize with strength, swing and groove templates, plus CC, pitch-bend and aftertouch lanes
Mixing: Group buses, FX returns, sends, and sidechain routing from any bus into any plugin's aux input
Automation: Read, touch, latch and write modes, on every fader, pan, send, group, return, master and plugin parameter
Plugins: Audio Units, CLAP and VST3 mixed freely in one chain, reorderable, with delay compensation
Tempo: Tempo maps, time-signature changes, tempo detection from audio, and varispeed playback
Arrangement: Named markers, a draggable loop brace, an arranger track, and saved track templates
Metering: LUFS and true-peak metering, spectrum analysis, and a tuner
Export: Mixes and stems as WAV, AIFF, FLAC, MP3 or AAC, with bit-depth choice, dither and loudness normalization
Control: Rebindable keyboard shortcuts, MIDI learn onto any parameter, and MCU, HUI and OSC control surfaces

What you'd expect from an NLE

Multicam: Live switching between angles, recorded as timeline cuts you can move, restyle or remove afterward
Transitions: Sixteen styles between adjacent clips (dissolves, wipes, irises, bars, slides, pushes and spins), rendered as previewed
Compositing: Position, scale, rotation, crop, opacity and colour per source, keyframed per clip for picture-in-picture and moves
Retime: Speed changes, reverse, ramps and freeze frames
Cutting: J and L cuts on linked audio-video pairs, ripple edits, and frame-exact snapping
Titles: Text rendered into the program output, with subtitles and tag substitution
Effects: ISF and OFX shader plugins, including parameters driven live by the audio mix
Export: Codec, bitrate, resolution and frame-rate presets, with audio and video resolved over one shared render range
Interop: FCPXML, OpenTimelineIO and REAPER project export

The numbers

Platform: macOS on Apple silicon. Linux is next.
Audio engine: Fixed 32-sample block, a 0.67 ms ceiling on every callback that can't be raised
Wire format: 48 kHz, 32-bit float end to end: the format TOAST carries between peers, so nothing is resampled, negotiated or converted at the boundary
Networking: TOAST (Transport Operated Audio Sync Tunnel): peer sessions over a shared clock authority with anchored transport
Timeline clock: 705,600,000 ticks per second: divides evenly into every common sample rate and frame rate
Plugin formats: Audio Units, CLAP, VST3: mixed freely in one chain, with delay compensation
Plugin latency: Latency Tolerance: a ceiling you set in samples. Anything over it is bypassed automatically, audio and video alike. Per-slot processing blocks from 32 to 2048
Import: WAV, FLAC, MP3, AAC, ALAC, Ogg Vorbis, AIFF / AIFC, audio from MP4 and MKV
Export: Mixes and stems, synced H.264 video, REAPER .rpp, FCPXML, OpenTimelineIO, Guitar Pro
Video: Multiple live cameras, GPU-resident compositor, audio-reactive ISF shader effects, titles and subtitles
Streaming: RTMPS to Twitch, YouTube, Facebook, Kick, Instagram, Vimeo and Restream
Control: Phone remote over the local network, MIDI controllers
Writing: Tabs and standard notation on the MIDI engine, lyrics with built-in vocal synthesis
Lighting: DMX over Art-Net and USB-DMX

Built on open work

JAMWire Studio stands on the shoulders of open-source software, open models, and open standards. Thank you to the people who made them.

Stem separation: Demucs by Alexandre Défossez (MIT)
Transcription: Basic Pitch by Spotify (Apache 2.0)
Beat tracking: Beat This! by the Institute of Computational Perception, JKU Linz (MIT)
Amp capture: Neural Amp Modeler by Steven Atkinson (MIT): the capture format and the A2 architecture our built-in trainer reproduces
Vocal synthesis: The worldline renderer and English phonemizer model from OpenUtau (MIT), built on WORLD by Masanori Morise (modified BSD), with the CMU Pronouncing Dictionary
Inference: ONNX Runtime by Microsoft (MIT), through ort
Plugin standards: CLAP (MIT) and the VST 3 SDK by Steinberg (MIT), through clap-sys and vst3-rs
Video standards: ISF by VIDVOX and OpenFX
Timeline: The 705,600,000-tick clock is the Flicks unit published by Facebook (BSD). Session interchange follows OpenTimelineIO
Audio and files: Symphonia (MPL 2.0), hound and Claxon (Apache 2.0), rubato (MIT), RustFFT (MIT / Apache 2.0), midly (Unlicense), quick-xml (MIT), serialport-rs (MPL 2.0)
Interface: egui and wgpu (MIT / Apache 2.0). The phone remote is built with Flutter (BSD)
Virtual audio device: libASPL (MIT)
Typefaces: In the built-in plugins: Ubuntu (Ubuntu Font Licence), Mr Dafoe by Alejandro Paul and Yellowtail by Astigmatic (SIL Open Font License 1.1)

VST is a registered trademark of Steinberg Media Technologies GmbH.