Skip to content

QA: ArSon Player M1–M4 physical ASIO + music/FX field acceptance #680

Open

Description

@masarray

Milestone B verified triage — 2026-11-17: P1 / player / qualification / field-fat. Classification ≠ completion of product feature or physical FAT.

Scope / release boundary: Physical music/FX ASIO test not replaceable by synthetic CI.

Exit criteria:

  • On current Windows build, verify player WAV/MP3/FLAC playback, KEY/TEMPO and A/B cue through one native ChannelId, BUS/MAIN and physical output.
  • Exercise music transitions while MIC FX live; capture device/driver/buffer, actual audible result, failures and reopen behavior.

Canonical issue ledger. Preserve historical evidence below; no speculative merging or rewriting approved audio.


M1–M4 live music field acceptance (hardware-only; not a code-development blocker)

Code PR: #676. Release-candidate source head: 0034fd5e5f20fe35900c95ae0868cc6148e8d0df.
Scoped real WAV/JUCE native test PASS.
Exact-SHA Windows field preview.

This is an operator/hardware acceptance checklist, not a fake ASIO claim from synthetic CI. Software CI covers decoder/queue/cue/transition invariants; physical validation needs a compatible external interface and live source.

Field matrix

  • ASIO (preferred) or supported WASAPI device with an authorized MAIN output: open a WAV/MP3/FLAC track, ensure music plays through ONE native PLAYER A ChannelId with existing FX, WIRE -> BUS/MAIN; no browser audio output
  • Hardware MIC monitoring/FX remains audible and undisturbed while player transitions (do not press Monitor to recover); test MAIN MUTE/SOLO behavior and selected physical port
  • Player KEY +3/-2 and TEMPO 80/120 with music; verify stereo quality by listening, not just synthetic frequency tests; default KEY0/TEMPO100 unchanged
  • Multiple setlists, rename, remove+undo, 2,000-song browsing/search; queue reorder and manual queue with Auto Next OFF; verify queued song is not removed after decode/routing error
  • NEXT with Cut / Smooth 180ms / Crossfade 500ms / 1500ms; repeat the same song >7 times; confirm no click/gap/underrun while MIC/FX runs
  • Song IN/OUT and Loop A–B across >10 repetitions at 80/100/120% tempo; disable loop, press Next, verify queue precedence and original source unchanged
  • Pause, seek, stop and WebView close/reload during live player; confirm native music transport and pending ACK remain correct
  • Device unplug/restart or safe sample-rate rebind between 44.1/48/96k, with pending NEXT; NEVER automatically resume audible output after device reconnect
  • 30-minute sustained music + MIC + FX test with device-specific ASIO buffer sizes; collect source PCM dropout/underrun evidence and any UI stalls

Exit evidence

Device model/backend and driver version, channel/port mapping, sample rate/buffer size, software SHA, actions attempted, expected vs heard result, and whether any underrun/logging is recorded.

Not in this work: full offline waveform cache, arbitrary VST3-per-song state snapshots, and sample-accurate no-gap certification. PR #675 addresses independent KEY/TEMPO DSP quality and requires separate subjective music A/B. Do not hold general native Player CI hostage to Vocal Harmony plugin tests.

Activity

  1. masarray commented on Nov 14, 2026

    @masarray
    OwnerAuthor

    Software engineering M1–M4 merged to main: b3ca20bafe08f2f62104d7f2df57bd4c723751a2 via PR #676. Focused JUCE Windows WAV regression: https://github.com/masarray/arsonmix/actions/runs/43555895602 (PASS). Final portable artifact: https://github.com/masarray/arsonmix/actions/runs/43556224841 (SUCCESS; ArSonMix-Windows-Field-Preview-0034fd5e5f20fe35900c95ae0868cc6148e8d0df). This issue now owns only physical ASIO/MIC/FX/music audition and performance acceptance; do not conflate with missing code implementation or Vocal Harmony CI.

  2. masarray commented on Nov 14, 2026

    @masarray
    OwnerAuthor

    M5 integrated source milestone merged to main via #682, exact main SHA 28cffcfc4ae595c1cf3b8447b79e0fc67cab2fb7. High-overlap worker-only KEY/TEMPO, smooth hot controls, two-sided musical EOF drain, and M4 cue OUT/dual-deck integration are implemented without changing player PCM authority/FX routing. Source checks on PR head d03fb5a65e96691c4b07c961b8f19b3bad2b6bf7 all SUCCESS: DSP, M1–M4 portable show, fast native, Windows JUCE real WAV + M5 cue/OUT. Keep this QA issue OPEN until physical ASIO + MIC/FX/MAIN and genuine full-band music KEY/TEMPO listening/soak evidence is obtained. These were not tested and are not release qualification claims.

  3. masarray commented on Nov 14, 2026

    @masarray
    OwnerAuthor

    M6 source milestone now merged: PR #684 → main@baf88dddfb789080c5348fa728b07aa4ed53bdd0. Approved Player UI now virtualizes 2,000 songs, displays all 128 queued entries, supports drag/keyboard queue reorder with native revision fencing, and avoids retransmitting unchanged 2,000-song catalogue on queue-only updates. Exact PR source 6dadd1bd43047e1992e36dc3c594d65034f1cbbf: Player native C++ + executed UI behaviours, Chromium, fast native contracts all SUCCESS. No new physical ASIO/MIC/FX test or exact M6 Windows MSVC/installer release was performed: keep this issue open. Complete waveform cache and measured gapless still deferred to the next source milestone.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions