Start With a Groove.
Finish the Record.

47 instruments from 52 packs: grand piano, a string section with sustained, pizzicato, spiccato and tremolo, brass, woodwinds, basses, guitar, a recorded kit, percussion and mallets. A 117 MB core set installs with the app; the rest download when first picked and stay offline.

Start from any of 18 presets: up to four oscillators (FM, a plucked string, wavetables built in or loaded from a WAV), unison of seven, a filter with its own envelope, drive and four LFOs, free or in time. A key played live is the very note the song renders.

16 pads, each with layered sounds, pitch, decay, choke groups, a hall send and its own effects, and a step sequencer. The 12 genre kits (88 sounds) come from the Drum Synth, so they're original by construction.

Any recording becomes an instrument: start, end and loop with a crossfade, the root note found by listening, an envelope and filter. Chop by hits, a grid or the beat onto pads, across the keys or onto the timeline; warp it to the song's tempo with its pitch kept.

Up to 32 instruments on one track, each with its own octave, level, pan, detune, timing, tone, swell, grit, drift and glide (the 808 slide), plus a selector that thins a stack in the verse and opens it in the hook.

Distance darkens a part and sends it further into the shared hall at once; every part shares a small room; width keeps the low end in the middle. Distance can move over the song.

EQ, compressor, reverb, plate, delay, chorus, saturation, limiter (true peak), multiband, filter, width, transients, doubler, lo-fi, gain, tape stop and stutter, plus ducking under the kick. Each shows its latency and how long its tail rings.

Up to eight knobs a track (Warmth, Width, Punch, Space…), each moving several settings at once and automatable, and a plain-words explanation of how any part is made.

Scenes and slots beside the timeline, launched on the bar, with follow actions, trigger, gate and toggle, pads that launch slots, and the performance recorded into the arrangement.

MIDI keyboards (MPE too), the computer's keys or the on-screen keyboard play any track live, 64 voices. Record several tracks at once, punch in and out, loop takes and comp them. A take survives a crash.

Chords from the key, strum, arpeggiate, fold, legato, move into the scale, quantise with swing, humanise with a seed. Lanes with curves and LFOs on any setting, exported exactly as heard.

Sing or play notes one by one and they become an instrument kept in the song, each note tuned to its pitch. The vocal range finder names your range and voice type, and a voice can be reshaped into another range (in the engine; the app's screen comes later).

Instruments and effects written in Rust on the shared audio core, so they run the same on every platform and the same in the browser build.

Sent through Waverate's artist service with your sign-in, splits and credits filled in. The engine never sees your sign-in token.

Audio, cover, release.json and the upload form's checklist with copy buttons, until the planned partnership gives a direct link.

WAV (16 and 24-bit) and FLAC at 44.1 or 48 kHz, with the ISRC embedded (ID3 in WAV and MP3, Vorbis in FLAC), a loudness report and cover checks.

Once Avlory has its registrant code, Sound issues codes from a registry that never hands out a number twice, even after a deletion.

The song, every edit and every recording are encrypted with a key that's only in the link, so the testing server keeps only scrambled bytes.

Recordings go up in chunks in the background. A cut-off upload sends only what's missing, and arrives bit for bit.

A desktop and a browser join the same room either way round, with per-person undo and who's here.

One shared audio core: a Voice recording drops into a Sound song as a track, copied in with a link back.

Release straight to Waverate with credits, ISRCs and splits filled in, through Waverate's artist service.

A ready-to-upload package (audio, cover, release.json and the upload form's checklist) until the planned partnership gives a direct link.