Member Avatar for HiFiDr

I and many others have found that when converting to and converting from FLAC format from/to WAV format there is a sound quality degradation even though file size remains the same and accurate rip comparisons in EAC show no differences. Performing these conversions on two different PC computers of differing OS (Windows 7 64 bit vs Vista 32 bit), differing RAM (4 vs 8 GB), and differing CPU's (2.66 GHz quad-core vs 3 GHz duo-core) indicated that the effect was present in both systems but smaller in the better computer, thus indicating processing power was one source of the problem. We also found that differing software programs also contributed to the problem. Finally, by repeating WAV to FLAC to WAV conversions 5 to 10 times, it appeared that the files themselves were somehow cumulatively degraded in sound quality. We have been told that these results are impossible since FLAC compression is lossless. We have no reason to doubt that this is true yet we have repeated our listening tests several times with different listeners and with strictly single-blind procedures and find our results to be reproducible. We suspect that jitter is the explanation but question how this can be propagated through several WAV to FLAC to WAV conversions. Can any Guru out there explain our findings, please!?

  • 2 Contributors
  • 3 Replies
  • 298 Views
  • 4 Months Discussion Span
  • Latest Post Latest Post by dannyniu

Dani AI

Generated

Quick expert summary and a focused checklist to separate “file changed” from “playback/psychology” causes. reported repeatable audible differences after WAV→FLAC→WAV cycles; sensibly asked whether the WAVs and codec were what they appear. FLAC itself is a lossless container, so the first task is to prove — with bit-level checks — whether the PCM samples actually differ. If the decoded PCM is identical to the original, the audible change must be coming from playback, resampling, dithering, normalization, or listener/room factors rather than file corruption.

Basic, reproducible checks (binary/hash compare and FLAC-tools):

# Windows (cmd / PowerShell)
fc /b "original.wav" "roundtrip.wav"
certutil -hashfile "original.wav" SHA256
certutil -hashfile "roundtrip.wav" SHA256

# Linux / macOS
sha256sum original.wav roundtrip.wav
cmp -b original.wav roundtrip.wav

# FLAC utilities (if installed)
metaflac --list file.flac    # inspect STREAMINFO (contains MD5)
flac -t file.flac           # test FLAC integrity

If hashes/cmp show differences then the conversion pipeline is altering samples (bit-depth truncation, resample, dither, clipping, or a buggy tool). If files are identical, shift focus to the playback chain.

Playback / perception checklist:

  • Force bit‑perfect output (WASAPI Exclusive, ASIO or equivalent), disable Windows “Enhancements” and any per-player resampling/volume leveling/ReplayGain or DSP.
  • Test the same pair of files through the same DAC/player to avoid driver/clock/jitter differences between runs. Jitter is a timing/playback effect; it cannot be stored in files and therefore cannot accumulate across file generations.
  • Re-run single-blind ABX tests with the identical playback chain and fresh random ordering; small room/level or fatigue effects can produce repeatable percepts if the chain changes between comparisons.

If binary differences are found, post the exact toolchain (encoder/decoder names and full command-line flags) and sample format (bit depth, sample rate). That information usually reveals whether an encoder or wrapper accidentally resampled or downsampled during one of the conversion steps.

Member Avatar for dannyniu

Are you sure you've used flac codec?
It's probably some other lossy codec supported by .flac files.

Please check and redo the experiment,
and if the result come up again,
please think about and prove why it is

Member Avatar for HiFiDr

Dannyniu,

One thing I forgot to mention, we consistently used the 0 compression setting for minimum compression possible (0 does not equate to no compression, just the minimum software setting). We, and now many others have found similar results and a good part of the problem undoubtedly lies with the background processes running in the computer. However, the persistence and increase in degradation occurring with multiple generations of files is quite a bit more difficult to explain. As far as whether we were actually using the flac codec, I really don't understand what you are saying. We went to the appropriate website, downloaded the flac codec, installed it, used it. In addition, the various software programs we tried all indicated the flac format. What did we miss and why wouldn't using any lossless compression codec be....uh....lossless?

Member Avatar for dannyniu

It's just my standard process to check and see if every step of mine is going right, and I was actually shocked to see this.
Did you went to this site:http://flac.sourceforge.net/, it's the only proper place because, you're not sure if you will get the right thing if you go to another website.
Are you sure you used PCM encoded WAVs as test, because WAVs support many other codecs in addition.
What was you testing sound?
And which part of the testing sound is "damaged"? Is it louder lighter, or the phase is smaller or bigger?

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.