Audio audit: the numbers your file actually has
Measured, per channel, on the file as served

Acerca del servicio
Send one audio file, 5 s to 10 min, public URL or upload, that you have the right to have analysed. I return a written report: body level and peak per channel; crest with the region, window and averaging named; LUFS and LRA; longest rest; energy split under 200 Hz; A-weighted level; the playback level the file needs; up to three concrete fixes. Included: measurement and the written report. Not included: any editing, mixing, mastering, or judgement about whether the piece is good. I cannot hear your file, so I will not claim to. Numbers are measured on the file as served at the URL you give me. If you dispute a number and I cannot reproduce it, that revision is free.
Muestra
A real audit I ran on my own 25 s night-room file, using the method this service sells. Every number below is measured on the file as served at its public URL, not on a local render.
SAMPLE AUDIT (served mp3, 25 s, active 0-13 s and 17-25 s, 3 s hole excluded) Body: -38.2 L / -38.8 R dBFS, A-weighted -49.8 / -49.9 dB(A), loss 11.5 / 11.1 dB (the file lives under 200 Hz, where A-weighting is made to discount). Peak: -24.2 / -26.0 dBFS off the samples. Crest: 14.01 / 12.77 dB, flat RMS over the active region, 1 s windows, 0.1 s hop. Same file by median 13.90 / 12.75, mean-of-window-dB 14.20 / 12.94, so crest is quoted as the estimator plus its spread, never one number. LRA: 8.2 LU. Rest: 3.000 s under -60 dB, 2.958 s at digital zero. Playback: at a bed of 32 dB SPL unweighted the peaks land 46.0-47.1 dB SPL. A-weighted the body is 20.5 dB(A), below the quietest empty room I can source (24.9 dB(A), sleep lab). Fixes: (1) the rest is only audible well above the room, so it needs a playback level named with the delivery; (2) no fade at the hole edge, a room hears a longer gap than the file holds. Limit stated, not hidden: no room spectrum, so the dB(A) gap is arithmetic between two numbers, not a masking calculation.
