Repository navigation
Conversation
…e and rhythmic echo DSP
… schema-versioned state
…g Windows CI target
… ChannelId engine
…nsitivity and native THROW ACK
… bypass and RT heap safety
…ompact desktop/mobile UI
|
/arsonmix-ci native cce6323 |
…on device reprepare
|
/arsonmix-ci native 39ba9a4 |
M2 exact-native checkpoint / follow-up regressionVerified on Windows Actions #42498186748 for
Post-green hardening in new head
Fast contracts + targeted Chromium on final head pass. Exactly one owner-authorized Windows native final-head run #42499813920 is running for new CTest. No additional full build is requested. PR remains draft; physical ASIO/singer/xrun and sample-accurate beat-phase alignment are not certified. M2 is stacked on draft PR #469: do not merge out of dependency order. |
|
Repository housekeeping (2026-11-10): Close as superseded, not merged. The M2 tempo implementation was included in the reconciled M1–M11 delivery in merged #497/#499. This older stacked draft should not be merged again against newer main. Remaining verified-device and musical-sync acceptance stays in #473. Original branch/PR history remains available for audit. Current main and any pending candidate must pass their own exact-SHA CI and hardware gates. |
ArSon Vocal Throw M2 — real native MUSIC ChannelId tempo feed
Closes part of #473. Draft engineering qualification, not field-certified release.
Dependency / stacked change
Built on top of still-draft M1 PR #469 (exact green MSVC/63 CTest baseline
4db384851e6f6378cca4fd037b744c98bd1f0ba3). This draft temporarily targetsmainto use the repository's owner-authorized exact-SHA Windows qualification lane; do not merge it before M1 is resolved. The M2-only diff is the comparison from M1's head to this branch.Actual native pipeline
Accepted
ChannelIdmusic/backing lane, selected explicitly from Vocal Throw editor -> native epoch/revision/role admission -> pre-FX tap from authoritativeDynamicNativeSessionPublisher::render()-> bounded 10ms mean-energy SPSC, no separate audio callback -> CONTROL-only onset correlation + BPM lock/fallback -> one precomputed sample interval per active VocalThrowProcessor. Audio retains dry zero-added-sample path; existing dual frozen tails keep their own intervals as music tempo changes.Delivered
Limitations / release gates
No duplicate audio engine or conflicting M4 routing authority.