Four hosts, and two real differences
To a Mac, Sonari is a MIDI keyboard that happens to be your hands. It turns up in all four with no driver and nothing to install. What differs between them is not the notes: it is who has to be told, and whether your hands can run the transport.
Sonari makes its own sound. It is a 64 voice instrument with 10,000+ sounds inside it, so it plays the moment you open it and before any other software is involved.
Sonari also sends MIDI. Lift a hand and a note leaves; open your fingers and the brightness leaves with it. Your DAW receives notes, bends and controllers, exactly as it would from a keyboard, and its own instrument makes the sound.
And in three of the four it can drive the session. Sonari offers itself as a control surface as well as an instrument, so a closed fist plays and stops and a pointed finger records, without a hand leaving the air to reach the space bar.
What never happens is audio. Nothing Sonari’s own instrument plays arrives in your session as sound. The wire carries the performance, not the recording of it. There is no plug-in to install and nothing sitting in your effect chain. If you want the sound Sonari makes, you export the take as a file afterwards.
Once Sonari has been wired into a DAW it goes quiet whenever that DAW is running, lets the host lead, and names whichever one is making the sound. The MIDI is unaffected either way. Ask for both and you get both.

Sonari being played on a Mac: a chord lands, the keys under it light as each note sounds, and the readout names the chord and the notes in it.
Sonari puts its own port on the Mac, called Sonari. There is no driver, no virtual cable to buy and no bus to create first. It is there from the moment the app is open, before the camera has found your hands, so you can wire the session up first and play second.
A second port called Sonari In takes traffic the other way. A keyboard or a foot switch plugged into your Mac can play Sonari, or start and stop its looper, which is worth knowing if you already have a pedal under the desk.
Two things are worth setting once. Notes arrive on channel 1, the bass hand on channel 2, and drums on channel 10, so a multitimbral instrument can be laid out to match. And MPE is off until you turn it on: leave it off for an ordinary instrument, turn it on when the thing you are playing reads a note per channel.
Sonari knows which DAW you are running and carries the steps for it, along with a test signal that plays a short arpeggio down the wire. Press it before you lift a hand. If the notes arrive in the DAW the routing is right, and anything you hear from Sonari itself proves nothing at all.
The least to do of the four, and there is nothing to switch on. Make a software instrument track, pick a sound, and Sonari’s notes arrive the way a MIDI keyboard’s do.
The one thing that catches people is arming. The track has to be record-enabled or selected, or Logic will show the MIDI coming in and play absolutely nothing, which reads as a broken connection and is not one.
Transport comes free here and nowhere else. Logic finds Sonari’s control surface on its own, so if it asks about a new control surface, say yes, and from then on your hands can start and stop the session and roll it into record.
For a per-note instrument, turn MPE on in Sonari and set the track’s MIDI channel to All. Every note then carries its own bend, its own pressure and its own brightness rather than sharing one set between them.
There is one more thing Sonari does only for Logic. A finished take can be handed over whole: a folder with the parts as separate audio files and a MIDI file beside them, with Logic opened on it. That is the one place Sonari’s own sound reaches your session, and it does it as files rather than down a wire.
Live does not listen until you tell it to, and this is the step people miss. In Live’s own Link and MIDI settings, find Sonari in the list of ports and switch Track on for the notes. Switch Remote on as well if you want the transport controllers. Then a MIDI track, its input set to Sonari, armed.
MPE has to be switched on twice. Once in Sonari, and once for Sonari’s port in Live, which has its own MPE setting per port. Miss the second and Live reads every note on one channel, which sounds like the expression has stopped working when it is arriving and being thrown away.
Transport is the part Logic gets for nothing. Live will not find the surface by itself, so add a Mackie Control surface in Live’s control surface settings with its input set to Sonari Control. If you would rather not, Live’s own MIDI mapping will learn the transport gestures instead.
FL Studio takes the same route as most hosts: enable Sonari as an input device in FL Studio’s MIDI settings, point an instrument at it, arm it, and play. For per-note expression, turn MPE on in Sonari and put the instrument you are playing into its own MPE mode.
Transport works the way it does in Live: add a Mackie Control surface with its input set to Sonari Control, and your hands get play, stop and record.
Of the four here, Sonari carries steps by name for Logic Pro and Ableton Live. FL Studio takes the general set, and the general set is the whole job.
Notes and expression work here as they work everywhere. Enable Sonari under Pro Tools’ MIDI input devices, arm an instrument track, and play into it.
The transport does not, and it never will. Pro Tools takes two kinds of control surface and Mackie Control has never been one of them, which is the language Sonari’s surface speaks. So there is no play, stop or record from the air in Pro Tools, and Sonari says so rather than offering you the setup. Everything else is untouched by it: the notes were never going through the surface.

Somebody playing Sonari with both hands raised in a studio, the shape of each hand drawn on screen and the pitch laid out along the bottom as they move.
The camera is the floor, and no amount of software moves it. Most built-in Mac cameras run at thirty frames a second, so on average about seventeen milliseconds go by before a movement has even been looked at. Reading the hand takes about three more. Sonari aims to start a note inside twenty milliseconds and treats thirty as the point where it stops being playable, which means the camera has spent most of that budget before anything else has spent a penny of it. A camera that offers sixty, an iPhone used as the Mac’s camera or a webcam that runs that fast, halves the first term and is the only thing that will.
Then there is your own interface, whose buffer adds whatever it adds after the notes have left. Sonari reads that number and leaves it alone: on a Mac, forcing a buffer size changes it for every application using the device.
Steadiness matters more than speed. A delay that stays the same is far easier to play than a shorter one that wanders.
A camera reads shape and position, and that is all it reads. It cannot feel how hard you pressed, because you did not press anything. A hand that runs out of frame runs out of notes. And a bright window behind you turns your hands into silhouettes, which is the single most common reason tracking comes and goes. Front light, chest height, about an arm’s length back, and room to move sideways.
The things a camera genuinely cannot sense have somewhere else to come from. Breath, the tilt of a wrist and a thumb on a surface arrive from an iPhone instead, alongside what the camera is already reading.
The traffic itself, message by message, is what Sonari sends over MPE. If you are still deciding whether a camera is the right thing to buy at all, the gesture MIDI controller page puts it beside the hardware, priced.
Sonari is an instrument you play in the air, on the Mac you already have. Nothing to plug in, nothing to set up, and it plays the moment you open it.