Skip to content

P3.5B: Capture and Sampling Lab - #7

Merged
astrobyte-dev merged 1 commit into
mainfrom
feat/p3.5b-capture-sampling
Sep 19, 2026
Merged

astrobyte-dev merged 1 commit into
mainfrom
feat/p3.5b-capture-sampling

Conversation

@astrobyte-dev

Copy link
Copy Markdown
Owner

Bring user-owned WAVs and external input into the existing Studio sample, FX and composition workflow. My Sounds now imports/deduplicates managed audio with immutable identities and cached waveforms; Capture provides explicit input/monitoring, validated dry takes and Keep as Sample. The Sample tab adds nondestructive trim, rate-based pitch, reverse, envelopes, phrase loops and stable chops/pad mappings, with shared Undo/Redo, persistence and structured MCP operations.

This draft is intentionally stacked on feat/p3.5a-sound-lab-core at 3ea7f7b4441bdd456c83699f9a170c9e6c9a56d9. Do not merge this or its lower draft dependencies as part of this delivery. P4/P5 were not started.

Validation

  • Baseline: 259 tests, 256 passed, three Windows-only skips; build, both typechecks, six browser journeys and Linux lifecycle passed before edits.
  • Final: 304 tests, 301 passed, three Windows-only skips; 45 new sampling tests. Production build and both typechecks pass.
  • All seven classic/Studio/P2/P3/P3.5A/P3.5B/System browser journeys and Linux runtime harness pass. Final display adjustment was followed by build/typechecks, 45 sampling tests and a repeated P3.5B journey.
  • Production Studio/HTTP/storage plus actual MCP exercise import/folder filtering/dedup, shaped previews, grouped trim history, FX, fixture capture/finalize/Keep, assignment/chops/sequencing/scenes, stale write rejection and complete save/reopen. Screenshots reviewed at 1280/1440/1920.
  • Security/path, destructive file mutation, audio ownership, cache/memory bounds and full diff reviewed. External originals and retained sounds are not deleted by musical Undo.

Acceptance limits

Implementation complete; native input/audio acceptance pending. Ubuntu evidence uses explicit deterministic audio fixtures, not a physical microphone or real SuperCollider/Tidal sound. Windows is unavailable. Native device/channel behavior, monitor/preview/output isolation, five-minute cutoff/finalization, sound/timing/CPU and buffer lifetime still require hardware acceptance.

All inherited gates remain open: P2.6 Windows desktop/audio; P3 native audio/timing and Windows; P3.5A native audio/timing/CPU and Windows. No system packages, kernel/apt state or native audio installation/configuration were changed.

Initial scope deliberately supports PCM16 mono/stereo WAV imports (256 MiB/15 minutes), one of two mono input channels, dry capture after gain (five minutes), track FX after Keep, playback-rate pitch and phrase retriggering. It does not offer live input FX, processed capture or elastic time stretching. Projects reference managed audio/ files; copy that directory/sidecars with projects. Missing files require exact-content relinking.

Full architecture, file inventory, test evidence, screenshots, exact exit criteria and remaining limitations: P3.5B completion report, validation JSON, task.

Base automatically changed from feat/p3.5a-sound-lab-core to main September 19, 2026 01:07
@astrobyte-dev
astrobyte-dev marked this pull request as ready for review September 19, 2026 01:07
@astrobyte-dev
astrobyte-dev merged commit 55108e0 into main Sep 19, 2026
2 checks passed
@astrobyte-dev
astrobyte-dev deleted the feat/p3.5b-capture-sampling branch September 19, 2026 01:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant