What is your preferred music file format and why?

Going back to the original topic, I find it interesting that of the compressed audio choices, the only two codecs that we even consider today for lossless compression are FLAC and ALAC. With FLAC as the basic standard everyone should support, and Apple lossless what you use if you want compatibility with an iPod or iTunes. And though they are different they are at least functionally the same for the common use cases.

I remember a time when there were all sorts of competing audio formats for lossless compression, with discussions and debates about which was best from several perspectives. Encode time, decode time, file size, and so on.

• WMA lossless from Microsoft. Not open source, but reverse engineered. But no one uses Microsoft music products on purpose in 2024.
• Monkey's Audio (APE). Still exists, open source. But I've never known anyone who uses it over FLAC.
• WavPack, which is also open source. It seems to handle more than just PCM audio, but again I don't know anyone who uses it. Maybe the DSD crowd has more experience?
• MPEG-4 ALS, which I don't think ever gained any use.
• La (lossless audio), which no longer appears active
• OptimFROG, which is not open source that I can tell but does appear to have a permissive license.
• Shorten, which I think is mostly gone at this point
• TAK, which I remember reading about way back in the day
• A few others I've forgetten, I'm sure.

But today just the two, and it would probably be the One if Apple ever just got over it.

 

That's a tradeoff, isn't it? It's also wasteful to store a copy in each format of everything in their catalog. A lot of titles have very low traffic and it's reasonable to use one format as a master copy and convert on demand to other formats.

 

I started ripping to .wav back in the day when hard drives cost about $1 a GB (2005?) but then my ass got bit a couple of times from iTunes update tomfuckery because iTunes would confuckulate something something...metadata. Then I ditched iTunes and tried Amarra and Audirvanna or some shizzle and went to FLAC. Also so I could copy to SD card and play stuff in the Jeep. But since 2018ish I either stream FLAC from Qobuz via Pi2AES for home or phone in the car. I pretty much ditched trying to maintain hard drives full of files to either streaming Qobuz, or if unavailable I buy used CDs.

 

@scblock showed with his analysis, the files are identical. There are only two rational conclusions I can come to. 1. Something is wrong with the decoder (highly unlikely) or 2. Expectation bias, or to put it less diplomatically, you hear what you want to hear.

 

Diplomacy? On this forum? I am aghast

More seriously, I understand having "these are two different file formats so they have to sound different" stuck in my head, but I feel like that'd not explain the ABX streaks where file IDs are masked and all I'm seeing are Buttons A, B, X, and Y. Granted, I can only do maybe 5 or 6 in a row which does imply lower probability, but I'm a really bad test-taker in general and I really don't have the critical listening chops that a lot of folk on here have; "obvious" things to others take focus on my end to catch.

I'm gonna install PlayPCMWin and see if that makes any sort of difference, thanks @Biodegraded! This looks cool.

For reference my foobar settings are essentially: all DSP off, WASAPI (push) into whatever DAC I'm using with a 1000ms buffer length. I used to have a SoX resampler installed but I've foregone that for years-- I'm also on a fully fresh Windows install as of ~two months back, something I do every couple of years cuz system responsiveness starts sucking after a while.

 

Only tangentially related, but any old school computer gamers here remember editing your own .mod files in your games?

 

It's probably a dumb waste of resources to zip a FLAC file!

I really appreciate the analytical skills shown on this thread. Juggling contents of binary files is not easy.

 

Since a difference is being heard on playback, how about digitizing the analog output and compare those files?

 

I think it would be pretty close to impossible to get identical files by recording an analog playback of them, though you might get audibly indistinguishable files if done well. Any tiny analog fluctuation would find its way back through the recording chain. Feel free to try, but I'm not set up to do it here.

Heck I couldn't even get two short digital recordings of the same file, every sample aligned and visually the same, to perfectly match when examined as I did before. I recorded with Audacity through the windows loopback device which does put the system mixer in play. But if I invert one recording and mix them down to a new track I get nothing even at the most extreme zoom levels. So call it "effectively" the same audio, but it's not bit-perfect in the file itself.

 

Just a friendly disagree because I don't think it works like that. If you have the time, patience, gear, etc, you could always try. But I'm guessing that it'd be a big job because one should show repeatability.

As I've confessed before: I much prefer talking about this stuff to doing it

 

I hesitated posting this because I lack self confidence and I cannot find a way to do a proper ABX with this, but out of the same PC but running PlayPcmWin instead of foobar2000 to at least try and eliminate that from consideration, I think the WAVs sound more expansive and nuanced still based on listening tests I did a couple dayss back; with my nature I am more inclined to think that that is pure placebo and that it really ought just be foobar messing up somehow.

I know it's an annoying ask, but I would appreciate others try to determine whether there's an audible difference between the files I shared, again both of which are direct downloads off bandcamp in their respective formats and are found to be bit-identical and null when downmixed after polarity inverting the one. Been slightly stressed out by other things so I lack mental wherewithal to be bothered by this.

I nuked all my Bandcamp FLACs over the weekend, replaced them with the WAV downloads, and then converted them to level 8 FLACs which now populate my library. I'm fine leaving it at that.

 

A few thoughts!

• Different levels of flac compression impose different CPU loads to decode as well as encode. The max level compression uses quite a bit more. On a PC or something that will get lost in the general emi shitshow but on a low power embedded device it's just possible you could hear a difference due to how this affects the interrupt timing that delivers the data blocks to the spdif/i2s output. If given a choice I use the medium flac compression but I don't care enough to reencode anything or to download wavs and encode them myself.
• Playpcm is good on PC. Mainly I like it because loading an album into memory takes a while, and this "cueing up time" is a disincentive for album hopping after one or two tracks.
• These digital tests lead to the most severe abx brain worms it's possible to get in audio. I have definitely driven myself mad a few times with them. I generally now just pick my instinctive first preference from whatever are the least hassle options and try not to think about it again.

Also wanted to answer rd's question here:

For upnp you have a server and a renderer. The server runs on the computer with all the files and controls the ways you can browse the files, the renderer (pi2aes or Holo red) just navigates the hierarchy the server provides. The server on my Synology NAS lets you browse by folder, and that's what I do exclusively. I imagine most servers will let you do this, but worth checking.

 

I was thinking I could just plug a hard drive into the Pi2AES USB ports and BubbleUPNP would find them as "local files" or in one of the other categories. Do I need a dedicated NAS with wifi? Thanks for clarifying!

 

I don't know for sure because I haven't tried it, but my understanding is that "local files" would be local to the tablet/phone you are running bubbleupnp on. If you can attach your USB hard disk to an android tablet that would work and you could browse the files on the disk and send them to the pi via bubbleupnp.

There is another way of doing things so that you can just attach the disk to the pi. You plug it in, add the disk as an extra source in moode/volumio, and then access via the moode/volumio web interface rather than bubbleupnp. It's less convenient because you have two different apps, one for streaming and one for local disk, but I think it should work.

 

Thank goodness I have poor hearing and can not discern anything above 256 Kbps MP3. Which has surely saved me a lot of memory space and heartache (money & time) as I (due to my OCD prone nature) would surely only want .WAV files... again, that is if I could discern a difference between uncompressed and lossless... between lossless and lossy... let alone between 256 and 320 Kbps MP3.

 

I wish everyone was as honest and humble as you are.

 

FWIW, I used to have a USB SSD plugged into my Pi2AES/Rpi and moode used the SSD as the local files... I used bubble to stream qobuz to moode and the moode interface to play the local files. I've since moved away from both the SSD and Bubble, so my fuzzy memory can't provide any more details.

 

Actually, I found 128 to be good enough. And that's what I use on the phone/etc.

And this was when my hearing was not nearly so crap.

As an aside. not having to reinstall Linux is part of the authentic experience. That whole reinstall culture was/is a Microsoft-crap thing.

But even Linux (on PC, at least) is not a real-time operating system because the underlying hardware is not real-time but time slicing/sharing. It cannot be perfect, and I do sometimes find that a re-boot is necessary for perceivably better sound.

 

I use MP3 quality off Qobuz whenever I'm in the car or on the move. In those cases ultimate quality is irrelevant due to road noise or doing chores or exercising or whatever. Also if I'm working and just mindlessly browsing Qobuz to discover new music.

If I'm at my dedicated rig purely for music listening, the difference between MP3 and ALAC is obvious and I can't do MP3.