What is your preferred music file format and why?

I'd like to say that I'm generally a very nonconfrontational person and I regret bringing this up to begin with because it's making me uncomfortable now cuz I feel like I have a target on my face or something.

Just to be emphatically clear, when I alluded to FLACs taking more processing power to decode than WAV, I was poking fun at a common misconception. I do not genuinely believe that would make a difference.

 

Don't worry about it, you're fine, I think it's an interesting experience. Have you tried the comparison with a different player, and not Foobar, or even a different computer?

 

I don’t know what to say, it’s wild and had me checking everything again today including looking at the WAV specifications.

Please, anyone feel free to double check my work here.

 

I think it’s fascinating. Why are people hearing differences in practice? Why when I convert the data it is the same? What else is going on? I just worry we junked up RD’s thread a bit.

 

I guess I am an intransigent old codger, I started out with .Aiff and never looked back. Storage got cheaper faster than Foobar could rip my discs, so I never worried about compression. Sure, lossless means lossless, I get it, but then you could just not compress at all and not have that argument.

 

FLAC compression setting wouldn't matter – all settings produce a lossless result. Higher compression just runs slower because it does more processing to analyze the music stream.

 

I get that, but maybe the media player and computer are having issues with the decompression, a setting maybe conflicting with the playback.

 

More likely: buggy code. Having been in the computing-with-complex-and-sometimes-realtime-data business for all my career, I could tell you many horror stories, from buffer overrruns to hardware glitches. Except that most are proprietary...

Here's a good one that is not propriaetary . Some years ago, there was a storm in the Roon forum about glitches when playing from Roon to various DACs over USB. Roon staff were berated for months. Eventually, the problem was that the reference USB receiver code from XMOS had an obvious counter bug (maybe wrap-around, can't remember now), and many USB DAC vendors had just taken that code and burned it into their firmware with no code review or testing. I'd totally not be surprised if similar things were happening with respect to reference implementations of FLAC decompression. A big benefit of open source is that it can be read, studied, and tested with access to the source. But only if that actually happens, which may not be in the budget of hard-pressed audio software and hardware vendors.

 

Can you extrapolate on this?

 

no way you made it much more interesting and useful! This is all useful for deciding which file format to use. Carry on!

 

Haven't tried another media player yet, I'm a creature of habit and while I do sort of recall thinking JRiver sounded better (cough cough, mind I trust my own impressions from thenabouts only scarcely) they're asking a LOT of money for playback software. Then again, seems dunking thousands of dollars into plugins is common practise for mic folk so I don't see why this oughtn't make sense. Might see if I can sneak another free trial in just to confirm it's not a foobar-specific issue.

By the by, ABX results with another Bandcamp album (not quite 95% confidence but I HAAAAAATE critical listening):


I even redownloaded the same Bjork album in FLAC and WAV to redo the test from scratch and make sure the first download didn't just get bollixed somehow. Same results. What I'm curious to do now is compare the first FLAC download which I have in the GDrive still (nuked it on my PC out of annoyance) with the more recent FLAC download which I got yesterday and check if I can hear an audible difference between those two.



One thing I notice on Bandcamp is that when you choose to download an album you don't just get a list of formats to choose from, you have to select a format from a dropdown menu and wait while it "prepares" your file for download. As you can see above Bandcamp music comes packaged in .zip, so maybe that's what Bandcamp is doing while it prepares files. I SINCERELY doubt it's that they're converting stuff every time, that'd be a remarkably dumb waste of resources. It can't be the zip-packing messing stuff up either cuz then the EVERY format would be affected, no?





Ugh I'm gonna try another player over the weekend. I HAAAAAAAAAATE critical listening.

I was chatting with someone over on Discord who's a lot more objectivist-leaning than a lot on here and they had me put both the FLAC and WAV in the GDrive into Audacity (which I do use to record stuff fairly often), invert absolute polarity on the one, and then export. I got -infinity on the latter which would imply they're the exact same:




And now you all know why I think I should be full of BS right now but nah, they are inexplicably audibly different somehow.

 

There's <something> in the playback chain awry. Exclusive vs windows mixer, DSP, resampling, replaygain, etc. that is different WAV vs FLAC.

 

$70 for JRiver is relatively cheap compared to most premium music players, and once you pay the $70 you can get the annual upgrades to the new version for $30 a year. Or if you love it as is, never spend another dime on it.

That is to say if you like the sound of JRiver I think it’s worth the cost.

 

I think this got lost in all my posts but I did convert the Bandcamp WAV to FLAC and was not able to discern a difference ABX-ing those two. Really think it's mainly Bandcamp screwing up with their FLACs somehow.

 

Ok let’s back up a second. You @Lyander and @ChaChaRealSmooth hear a difference between the FLAC and WAV files of the same album downloaded from Bandcamp. But you don’t hear a difference between your own FLAC conversion from that WAV file.

But before we scream bandcamp -gate @scblock looked at the same files you both heard a difference between at a bit level and they were the same.

There’s a piece missing here that connects both your impressions and @scblock ’s findings.

 

It’s like wav but can store metadata tags like flac. Populated by default on bandcamp.

Exactly, I can pick any format, and I just so pick AIFF on bandcamp, its like wav but its a little bigger, sure I can go alac or flac but I dont worry about it on most devices. Unless I have a old ipod, iRiver, or something else. A small format doesn't appeal to me for archival storage. Similar for Bleep, I just go with the format that has the metadata that I can.

If I rip a CD its in wav with a m3u file right besides it.
I don't convert all my files to be the same. I just get what is best for myself.

 

Maybe more people could try ABXing the files in the GDrive link on page 1 and reporting back here because I frankly think I'm well off my rocker in some respect or other (and that ChaCha isn't much better ).

Joking aside, I will say that Dan seems to be having a much easier time of this as I really do have to focus to catch the differences because they're subtle to me. Again, I got some new components in because someone was shockingly generous and gave me free Schiit on top of what I just bought off them so I'm basically left recalibrating my baselines for how things "should" sound because I feel compelled to write impressions of the stuff that may (hopefully, really) prove useful to others before they spend money on a thing.

There's this blind mic comparison thread I posted for fun a while back. It was really easy for me to suss differences out with my Modi 3+ and Piety chain because I was listening on that daily for a bit over a year, yet when I swapped the Modi 3+ out for another DAC I suddenly had a lot of trouble differentiating them; I do have a BIT of faith in my listening skills but I'm kinda in the middle of a system recalibration which is why I harboured loads of doubt about this all at the beginning.

... Being familiar with a system does help but I don't wanna go back to the Modi 3+.

I do likewise suspect Bandcamp is somehow screwing things up because while I wouldn't necessarily say their FLACs sounded very obviously compressed (to my ears and at my level of listening discernment), they at least sounded different enough from the WAVs that I randomly noticed enough of a difference for me to somehow get within farting distance of 95% confidence ABX despite bad test anxiety.

And yet the files are bit-perfect based on @scblock investigation and my rudimentary polarity inversion test.

And then I managed to successfully ABX another random track off a different album I own (mind, I did explicitly pick a track that I felt would be easier to ABX, something with a lot of treble detail and nuance; you can stream it here or purchase the track so you can try different format downloads yourself for pretty cheap— I like their music a lot, very feel-good [thank you @Senorx12562 for introducing this band to me]: https://lydiancollective.bandcamp.com/track/legend-of-lumbar).

I think I should be wrong based on evidence and just common sense. I don't know how even one other person, someone else whose listening skills I respect, actually agrees at the sonic difference. Admittedly this is kinda fun, but it also feels dumb.

I blame Epic Games. I remember they purchased Bandcamp some time ago, and they ruin everything they touch. EDIT WAIT I FORGOT THEY SOLD BANDCAMP LAST YEAR. I blame the new owners, Songtradr, but mostly Epic still.

 

I decoded the same FLAC using the official FLAC command line tool (previously I used foobar to convert) on a completely different computer and got the exact same result. There's nothing wrong with the Bandcamp provided FLAC that I can see. The audio is identical to the WAV after decoding. SHA-256 checksums of the audio data are identical, HxD doesn't see any bit differences in the audio data, etc. I don't see how we can possibly claim Bandcamp is doing anything wrong if two separate decoders show the same result which is also the same as the source WAV.

I'm going to add some screenshots here that I hope make clearer the minor differences in the files I described before. The specific differences in the overall files are limited to the file length portion of the WAV header (verified) and the metadata at the end of the file. The audio data itself after decompressing the FLAC is 100% the same. The rest of this point might get kind of boring.

Here I've highlighted the 44-byte WAV header for both files. Top is the original WAV downloaded from Lyander's link. Bottom is the FLAC file from his link converted to WAV using flac.exe version 1.4.2. The command used was flac.exe -d "Björk - Bastards - 01 Crystalline (Omar Souleyman Remix).flac" -o bjork.wav



Here I've highlighted the only bytes in the header that differ. These bytes are the total file size less 8 bytes (I don't know why the less 8 bytes), according to the header documentation. The byte order is little endian. Original file reports 682F9104, or 76,623,720 bytes. Windows shows 76,623,728, so that checks out. Second file report 242E9104, or 76,623,396 bytes. Windows reports 76,623,404, so this also checks out. The file size difference is 324 bytes, which we'll come back to at the end.



The actual audio data size is the last 4 bytes of the header. I've highlighted that in this next photo. These are both 002E9104, or 76,623,360 bytes, which is the exact length of the remainder of the original file before the start of the metadata for the original, or the remainder of the file for the converted FLAC.



The next 76,623,360 bytes are the actual audio data. If I jump to any location in the file we can visually see the data are identical, though manual review of 76 megabytes is impractical. If we rely on software, then we can show they are byte for byte identical by either asking the software to highlight the next difference (it finds none) or calculate a checksum. Both files have a SHA-256 checksum of the audio data itself of E3D990FCC694870CFD57BE5A136778BAFF84FA24B5E221BABB549C9597A27A0E.



After this is the final difference between the two files. The original includes 324 bytes of additional data at the end in a WAV LIST block as well as some ID3 tags appended to the very end of the file.



I don't mean to harp on this too much but given this result I simply can't find any issue with the provided FLAC.

 

Yeah, this is starting to read like the Computer Audio Players thread, where differences between players that "shouldn't" exist are nonetheless audible to people.

Try PlayPcmWin in exclusive mode. It's free, lightweight, easy to install and delete, and reads the full track into memory before playback so might minimize any hardware-dependent effects of different decompression requirements or any similar such mysteries.