Skip to content

P1K: measured native LUFS-I and 4x+ dBTP DSP (no simulated audio readings) #135

Open

Description

@masarray

Milestone B verified triage — 2026-11-17: P1 / metering / feature / backlog. Classification ≠ completion of product feature or physical FAT.

Scope / release boundary: Display only genuine measured loudness/true peak.

Exit criteria:

  • Implement BS.1770 integrated LUFS with correct absolute/relative gating and channel weights, and ≥4× oversampled intersample dBTP.
  • Validate offline PCM vectors at multiple sample rates and streaming bounds; no fake sample-peak renamed as dBTP, no RT allocation/locks.

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


Product requirement (2026-10-29 screenshot)

MASTER small black glass readout must show only LUFS and dBTP; green normal, red if integrated LUFS > -12 or measured dBTP > -1. P0 GUI PR #133 implements geometry, colors, thresholds and honest browser-only demonstration. Native currently shows em dash because engine has no BS.1770 loudness or oversampled true-peak measurement. Do not label regular sample peak as dBTP or substitute a sine-wave demo for audio evidence.

Engineering definition of done

  • Implement independent reusable preallocated DSP analyzer for native MASTER PCM at output rate, ITU-R BS.1770-5 K-weighting, proper gated integrated loudness (absolute/relative gating), channel weights and disconnection/reset semantics. Distinguish LUFS-I from LUFS-M and LUFS-S.
  • Implement validated ≥4x oversampled interpolation true-peak estimator for intersample maxima; linear sample peak is not dBTP. Handle 44.1/48/96 kHz, zero/NaN/Inf, DC and signed clipping.
  • No memory/locks/disk, FFT planning, JSON or WebView calls in audio callback; bound additional callback CPU costs and allocations via deterministic tests. If test hardware cannot run analyzer in budget, gate it by explicit capability with no false reading.
  • Native status transports measured fields with provenance, valid/invalid indicator, update ≤12 Hz via existing control timer, renderer uses same two-row masterLufs/masterDbtp IDs and red threshold class. No extra GUI animation loop.
  • Golden-vector test for K-weight/gating and intersample peaks, sine/impulse/inter-sample overshoot, channel weights, time windows, reset, latency; MSVC CTest and physical WASAPI/ASIO listening FAT. Preserve P1J graph and P0 — Crackling with 0 VST3: diagnose WASAPI clock/input/output buffering before engine expansion #15 xrun diagnostic independence.

This is a separate milestone from the visual-only PR #133 and P4 native Shows work.

Activity

  1. masarray commented on Oct 30, 2026

    @masarray
    OwnerAuthor

    2026-10-30 code audit update (do not conflate old issue baseline with current HEAD): engine/dsp/NativeMasterLoudness.h now exists, and tests/native_master_loudness_test.cpp includes an actual generated PCM 48k K-weight/gated LUFS-I and reconstructed intersample true-peak test. At current main, prepare() reports supported=false at 44.1 kHz and activates only approximately 48 kHz; the issue body's statement “native currently has no BS.1770 measurement” was accurate when first reported but is now outdated. This does NOT mean full #135 is complete: verify actual live UI wiring, calibrated multi-rate measurement, reset/device epoch and independent physical hardware FAT before close. This issue remains OPEN and is linked as MS7 in the new LV1/Studio Pro roadmap. No fabricated LUFS/dBTP or false sample-peak labels.

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