ITU-R BS.1770 loudness via ffmpeg's EBU R128 engine — the same standard the delivery specs cite. True peak is oversampled, not sample peak.
Each file is checked against your target: integrated loudness within tolerance, true peak under the ceiling. Failures say which check and by how much.
JSON for machines, job report for humans. Every check lists the measured value so there is nothing to re-litigate.
| Check | Standard | Status |
|---|---|---|
| Integrated loudness (LUFS) | ITU-R BS.1770 / EBU R128 | live |
| Loudness range (LU) | EBU Tech 3342 | live |
| True peak (dBTP) | ITU-R BS.1770 (oversampled) | live |
| Noise floor & hum | — | roadmap |
| Phase / polarity | — | roadmap |
| Clipping & dropouts | — | roadmap |
| Sync drift (LTC / clap) | — | roadmap |
Targets: −23 LUFS (EBU), −16 LUFS (podcast/stereo web), −14 LUFS (streaming music), true-peak ceilings by spec. Tolerance adjustable per job.
A 10-minute short is two blocks. A feature is billed like a feature — not like a seat you forget to cancel.
| Plan | Blocks (5 min) | Price | Effective |
|---|---|---|---|
| Free | 1 / month | $0 | — |
| Standard report | 1 | $1.50 | $18 / audio-hour |
| Pack | 20 | $25 | $1.25 / block |
| Pack | 100 | $99 | $0.99 / block |
| Full QC (roadmap) | 1 | $3.50 | $42 / audio-hour |
Basic loudness pass/fail is free elsewhere; the paid part here is the batch pipeline, the detailed report, the API, and integration into your dailies flow. A 10-minute QC costs less than 30 minutes of an engineer's time.
No. Sound Lab measures and reports; it does not change your mix. Fix decisions stay with you or your mixer.
Audio files today (WAV, AIFF, and the audio track of most containers). Batch multiple files in one job.
Yes — loudness target, tolerance and true-peak ceiling are per-job options, so Netflix, EBU, and podcast specs are all expressible.
One free block every month while Sound Lab is in beta.