“Sounds the same” and “is the same” are different claims, and a lot of audio QA confusion comes from treating them as one. A listening test — even a careful, trained one — is checking the first. Bit-Exactness, one of the six checks on a WaveLens report, is checking the second, literally: is every sample identical, sample for sample.
A report that looks contradictory until it isn’t
Here is a report from comparing one 1 kHz tone generated at 24-bit with the same tone generated at 16-bit. Same signal, two bit depths:
Spectral Fidelity PASS Bit-Exactness FAIL · 230,000 of 240,000 samples differ Level Fidelity PASS Artifact Detection PASS Distortion PASS Loudness PASS Overall Verdict PASS
Five checks pass, Bit-Exactness fails on nearly every sample, and the Overall Verdict is still PASS. The 16-bit file’s samples are rounded onto a coarser grid, so almost none of them equal the 24-bit values — but the difference is far below anything audible, which the other five checks confirm. The Overall Verdict stays PASS because Bit-Exactness is non-gating by default. That isn’t the tool shrugging off a failure; it’s the tool correctly recognizing that “not identical” and “not the same audio” aren’t the same claim. A codec round trip, a resampler or dither will produce the same pattern.
Where the strict claim actually matters
The cases where you genuinely need bit-exact equality are narrower than they first seem:
- Confirming a “lossless” archival copy really is lossless. If a vendor or a pipeline step claims a transfer is lossless, Bit-Exactness is the only check that can actually verify that claim — any difference at all is the finding.
- Verifying a file survived storage or transfer untouched. A backup, a CDN cache, a database blob column — anywhere a file is supposed to come back exactly as it went in.
- Checking a build pipeline’s “no-op” path. If a step in your pipeline is supposed to pass a file through unchanged under certain conditions, Bit-Exactness confirms it actually did.
For any of these, add --bitexact to make a Bit-Exactness failure fail the overall verdict — otherwise it’s informational only, exactly as in the report above.
Where it doesn’t
Everything else — did processing change the sound, not just the bytes — is what Spectral Fidelity, Level Fidelity, Distortion, and Loudness are built to answer, with real tolerances instead of a strict yes/no. A codec round-trip, a valid resampling implementation, or intentional dithering will all break Bit-Exactness while leaving those four checks clean, which is exactly what the report above shows. See Bit-Exactness vs. Bit Accuracy for the tolerant middle ground when you want a measured answer instead of a strict pass/fail, and Reading a WaveLens report for how all six checks combine into one verdict.
See it for yourself.
WaveLens compares a reference file against a processed copy and shows you exactly where they diverge — down to the sample — not just whether they still sound the same. One command, one report you can point at.