# ByteBeats research

Full9694-character indexed script inspected. Important final override: playpause is declared twice; the trailing declaration sets playing=true and resumes sound, so the earlier toggle behaviour must not be described as effective pause. Touchend also activates. No explicit source.start call appears, though connected ScriptProcessor is used. Code analysed without independent execution or raw redistribution.

Token hash byte pairs consumed from end using pop/256; finite parameter choices not runtime random. Six weighted families melodic/heavy/wobbly/powerclimb/absurdities/permutations; crazy3 exists but weight0. Four visual modes with bytes appearing twice in selection list. Semitone-index expression gives effective rate rounded from3951.07*2^((o-6)/12). Main40-140byte claim concerns generated expression, not full9694character renderer. Sample functions constructed with Function after Math.pow/sqrt replacement. 8bit mask255 recentered via(value-128)/256, gain.25, AudioContext requested44100. ScriptProcessor1024samples connects through gain. Time advances only when playing in callback. VizSquare quantizes draw time to size/slow factor. Bytes brightness relates absolute distance from128 and transposed sample coordinates; bits maps binary planes; waveform connects amplitudes; starry plots points. Initial preview searches100samplepositions from0to99000(step1000), chooses highest sum of RGBA pixels (called density), not necessarily time0. This must be reflected in close readings. Metadata regenerates parameters; progressions regex only captures first digit after c=b%, so not automatically a musicological count. No proof all outputs loop/neverrepeat.

Primary ArtBlocks page read and Dadabots site read in own browser: compact synthesis compared with large neuralmodels, credits Viznut for popularisation and KyleMcDonald bitmap visualisations; those are attributed claims. Linked Viznut blog failed in web reader via Google restriction; no direct reading claimed. MDN ScriptProcessorNode inspected: deprecated, AudioWorklet replacement. Current site historical energy/perk language not adopted as current fact.

Own browser canonical#0initial frame inspected: intense purple field broken by black horizontal bands of varying lengths, highfrequency strips left and wider intervals right. Click tested with supported WebAudio diagnostics. Freshcontext suspended then running afterclick, requested44100; realtime currentTime only0.00580499 at check, no visible advancement across screenshots. Mac locked may affect audiooutput, but cause unproven. No listening access and no claim of having heard output. Need runtime/presentation verification before calling essay ready. #1 initial Waveform frame inspected: pink dense lower mass, tall left spikes and ascending diagonal array toward right. #10 orange Bytes frame inspected: repeated horizontal registers, coarse dark blocks above finer texture and middle divisions. #20 yellow/black Bits frame inspected: narrow bands repeated at left and centre, larger intervals to right. Held472 initial frame inspected: green/black pixel grids, small square columns, broad bands and paired horizontal regions. All are initial-frame observations; no audio or sustained playback claim.

Fresh indexed held472 owner matches AB5Dwallet0xdd19308a9731cdecb67e0d80ca6823394208c045. HeavyMelodic/Bytes/Electric/rate4699/progressions3. Currentindex512/512,firstApril2 05:47:06,second/start13:58:52,activation14:04:24,completionJuly26 18:52:47,currentprice0.24ETH. ReleaseApril2. NIFTY label (not NIFTY License exact), termsnull; artist-code licence unknown.


Draft prepared with eight material sources and four identifiable initial-frame readings. Publication remains pending sustained runtime/presentation verification. Not added to publication manifest or counted as complete/live.


## Runtime recheck, 18 September 2026 (Melbourne)

The original canonical renderer now demonstrated visual playback in an isolated Chrome tab. For #0, a click changed WebAudio from suspended to running at 44,100 Hz; real-time clock readings advanced from 5.67437641723356 to 10.298049886621316 seconds. Purple/black bands visibly changed across screenshots. For AB5D-held #472, the audio clock had reached 8.530430839002268 seconds before a click while the green opening grid remained stationary. After the click, clock readings were 12.66938775510204 and 25.46938775510204 and the picture changed into shifting vertical stripes and checker-like regions. This demonstrates why a running audio context alone is not evidence of composition playback.

The full cached 9,694-character original script was reread. It sets playing=false, feeds zero samples while inactive and advances its counter only while playing. The final playpause declaration activates rather than toggles. The renderer was not modified and no browser settings were changed. The diagnostic domain was used read-only. Research tab was closed after checks to stop playback. No audio was recorded or listened to, and no subjective sonic description is supported. The earlier frozen-context observation remains an environment-specific result with unproven cause; it must not be generalised as a broken artist renderer or attributed to the Mac lock without evidence.

Held token 38000472 was rechecked through the exact project/contract Hasura filter at 2026-09-17T21:33:53.300103+00:00. Owner remained 0xdd19308a9731cdecb67e0d80ca6823394208c045. Response is preserved in evidence.json. This recheck verifies the token owner against the previously identified AB5D wallet; it is not a new census of every wallet linked to the profile.

Publication exception narrowed: visual playback is established, but listening and final embedded presentation remain unverified. The user requested autoplay for animated works, while this original artist script deliberately waits for activation. The coordinator should preserve original behaviour and explicitly resolve that presentation exception rather than alter the artist program or claim a static poster autoplays. No new publication-ready status is asserted.

## Coordinator embedded review

The original remote renderer wrapper was inspected in the in-app browser. Click activation began replacing the green opening image with new columns. Original gesture activation is retained and disclosed. No listening claim is made. Complete essay token usage is unavailable and shown explicitly rather than replaced with repair-only telemetry.