Bytebeat vs 1874 Sakura

t*(42&t>>10) That one line is the whole song. Bytebeat turns a formula straight into sound. It evaluates your expression while t counts up from zero, one step at a time, and writes the result out as the waveform. No synthesizer, no audio files, no samples. The formula is the music. ■ Edit while it plays - Rewrite the formula without stopping playback. t keeps running - Change a single character and the song becomes something else - A keypad holding only the characters that appear in bytebeat formulas, so one tap is one character ■ See the waveform - An oscilloscope draws the samples that are actually playing - Traces are aligned to the rising edge, so the wave stays still instead of scrolling - Instruction count and stack depth are shown too ■ 29 famous formulas - Includes the one-liners from the experiments viznut published in 2011 - Save the ones you like to favorites with your own names ■ Three modes - Bytebeat: result & 255 mapped to 0...255, the way the original sounded - Signed: treated as signed 8-bit - Floatbeat: the result used directly as a -1...1 waveform, where sin and cos come alive ■ Change the speed of t - 500 to 48000 Hz. The original bytebeat ran at 8000 Hz - Slower runs lower, faster runs higher ■ Export to WAV - 8-bit, 8000 Hz, mono, which is exactly what bytebeat is - 10 seconds, 30 seconds, 1 minute or 3 minutes ■ What you can write - The variable t, and the constants PI and E - Arithmetic, bitwise operators, shifts, comparison, logic, the ternary operator - sin cos tan asin acos atan abs floor ceil round sqrt exp log sign int pow min max atan2 mod - Everything is computed in Double and rounded to 32-bit integers only for bitwise operations and shifts. This matches JavaScript's ToInt32, so formulas you find on the web play as they are ■ No connection - No account to create - No ads, no tracking - The app never connects to the internet. It works on a plane Many of the presets come from the experiments Ville-Matias Heikkila (viznut) published in 2011.

1874 Sakura Studio writes complete pieces of Japanese and Chinese music and plays them on koto, guzheng, shakuhachi, erhu and a taiko ensemble. Press generate and you get a whole piece in jo-ha-kyū form, the shape of nearly every Japanese performing art: a slow opening called by a temple bell or a pair of wooden clappers, a development, a still centre where a flute plays alone over the shō, a rapid finish where the whole ensemble converges on the melody, and a closing that thins back to nothing. One melody carries each piece. It is previewed, stated, slowed down alone over the drone, and lifted an octave for the finish — so you can hum it back after one listen. The harmony works the way this music does: no chord changes, but a tone centre that moves by fourth or fifth over a tonic the shō never leaves. Every instrument is synthesised from how it makes sound. Plucked strings — koto, guzheng, shamisen, biwa, pipa — are modelled as vibrating strings, with the koto's pressed-string bend. Flutes — shakuhachi, dizi, xiao, hichiriki — are mostly air, with the pitch arriving from underneath. The erhu slides and sings. Temple bowls and bells ring from inharmonic partials. The taiko ensemble plays from real recordings, or switch them off and every drum is modelled. The pieces come from chaos maths. A Lorenz attractor picks the melody's rhythm and shape and then its notes; a logistic map drives the percussion, so no two bars need be alike. A small model learns as it writes, rewarded for the fourths, fifths and steps this music is built on. The CHAOS / TASTE slider decides how much say each one gets. Twelve styles. Eight traditional — Sakura folk, Gagaku court music, Guzheng, a Taiko ensemble, Zen meditation, Silk Road, Kabuki theatre and Neo-Tokyo fusion — and four fast action styles built for games: Samurai, Shinobi, Wuxia and Cyber Edo. Twelve pentatonic and Chinese scales, from hirajōshi and in-sen to gong and yu. Choosing a style loads its melody instrument and its ensemble. Every piece is built from its seed. Write the seed down and you can have that exact piece again, note for note. A seed from the browser version rebuilds the same piece here. You can play over it, too. The pads are the notes of the piece's scale rather than a piano keyboard, so nothing you touch is wrong, and the sixteen-step lane follows the piece and your host. Both play the melody or any part of the ensemble. Turn on RADIO and it keeps playing, picking a new style, key, scale and tempo for every piece. You also get: • 12 styles, 12 scales, 12 keys, tempo and density • 24 instruments and 13 one-click ensembles • A full instrument editor: string, breath, bell, vibrato, filter and envelopes • Sampled or synthesised taiko, switchable • Six-part mixer with a meter on every part • Temple reverb, tempo-synced echo, ensemble width and a filter sweep • MIDI in, so your keyboard plays whichever part you have selected • Export a MIDI file, a WAV mix, or stems — one WAV per part (in the app; in a host you record the audio there) • A note and a MIDI channel for every drum • UI scale, so the panel is not tiny on a big iPad • Live spectrum and waveform, and the chaos engine drawn as it runs Two Audio Units, an instrument and a MIDI processor. Each part sends on its own MIDI channel and every drum has a note and a channel of its own, so one instance can drive a rack of your own sounds. The MIDI version passes your keyboard through. Both follow your host's tempo and transport, and one install registers both. The app is free and nothing in it is held back. Using the Audio Units inside other apps is a one-time purchase, and each AUv3 session plays for two minutes first so you can try them properly. No account, no ads, no subscription. Works offline.