matthew (go nuts, go wild)
Every sound is computed sample by sample in your browser. The only recordings are your own messages, read aloud by your Mac and chopped into the song.
Play along
Hit the pads with your keyboard or fingers. While the track plays, hits snap to the next 16th note. Pads glow when the song plays the same thing.
Put your voice in it
Record up to 12 seconds for the track that's playing. Leave small pauses between phrases. Your phrases replace the spoken samples in order, your longest word becomes the pitched melody, and the whole take opens the track as the voice memo.
Press the red button and talk.
What's in it
Three tracks, one engine: five stems, 31 mix buses, a voice memo each. No samples, plugins or libraries.
Drums
- Kick A sine swept from 320 to 47 Hz, a noise click, saturated.
- Next-room kick The same kick low-passed at 260 Hz, so the party is on the other side of a wall until the rise opens the door.
- Clap Three noise bursts 10 ms apart and a band-passed tail.
- Hats and shaker Six detuned square waves (the 808 recipe) plus filtered noise, swung 57% for UK garage shuffle.
- Break A crunchy tuned snare with ghost notes, layered under the second drop.
- Ride and hand drums Six square waves ringing for the ride; tuned sines with a skin slap for the congas in the soft dembow pulse.
Keys
- Felt piano Up to 14 inharmonic partials per note, two detuned strings, a hammer thump, tape wow and flutter.
- Supersaw Seven detuned saws per note, pumping against a kick you can't always hear.
- Glass pad Pure partials that each breathe on their own slow cycle. The calm track's bed.
- Chords Four per track, extended (maj9, 6/9, m11) the way his records voice them.
Voice
- Choir chops A formant synthesiser: glottal pulses through five vowel resonators, sliding between oh, ah and ee, with a scoop into each note.
- Your words Samantha, Daniel and Paulina (macOS text-to-speech voices) reading your messages, band-limited like a phone memo.
- Pitched words "wild", "this is amazing", "calma y hermosa": pitch-tracked, then re-sampled so they land on the melody's notes.
- Choir pad The formant voices held long across whole chords.
Bass and air
- Sub A sine with a little drive, so it survives laptop speakers.
- Rolling bassline A plucked saw an octave over the sub, on the off-beats with 16th-note octave jumps.
- Reese Four detuned saws under a slowly moving filter, in the second drops.
- Crowd Nine band-passed noise "talkers" wobbling at syllable speed, swelling into a cheer.
- Rain and birds Filtered noise with random droplets, and chirps made of fast sine sweeps.
- Crackle Hiss and random pops.
Signal path
- Instruments render into 31 buses
- Bus EQ, drive, phone filter, tape wow
- Filter automation per bus
- Sidechain pump on every beat
- Reverb: 8-line feedback delay network
- Dotted-8th ping-pong delay
- Stems: drum glue, mono bass, mono below 150 Hz
- Your mute and filter moves
- Master: tilt EQ, glue compressor, soft clip
- Look-ahead limiter at −1 dBFS
Where it falls short. Nobody mixed this by ear. Levels were set by measuring loudness and spectral balance in each section. The chops are a synthesiser imitating a singer, and the automation is typed in, not played.