FLAC Lossless Uncompressed versus WAV

0 Members and 1 Guest are viewing this topic. Read 24544 times.

Music Matters

  • Jr. Member
  • Posts: 20
FLAC Lossless Uncompressed versus WAV
« on: 3 Feb 2012, 03:53 pm »
I am relatively new to the BDP-1, so this is what I have done so far:

Bought new version of dbpoweramp and ripped my CDs to FLAC Lossless Uncompressed
Using SSD external drive
Using mpod on my iphone as control interface

So as I have read more, some questions have come to mind and I am hoping for some feedback:

1) Thoughts on audio quality between FLAC Lossless Uncompressed and WAV?

2) File size between the two formats is slightly larger with FLAC versus WAV. Anyone know why?

3) "Loss of metadata" and "WAV lack of metadata support" is what consistently appears in print that would seem to give the nod to using FLAC encoding. With dbpoweramp used to rip, and if I ripped to WAV, at what point or how is there "loss of metadata"?

4) With the BDP-1, is there any practical advantage to FLAC versus WAV using the various control interfaces: mpod, mpad, gnome, mini, max as far as how album art, song titles, etc. will display? I have only used mpod at this point.

Thanks very much for any replies. I am still learning!!

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #1 on: 3 Feb 2012, 09:53 pm »
1) Thoughts on audio quality between FLAC Lossless Uncompressed and WAV?
The end goal with both formats will be to provide a bit-perfect PCM stream that matches the CD.   It all comes down to how bug free the codecs are to encode and decode the intermediate file format.   Since WAV and AIFF are both older formats and they are bound to have less bugs than Flac:lossless.  Tag support is clearly a major advantage with flac and AIFF over WAV and FLAC takes this a step forward with it's checksum utility on the actual PCM data.   If AIFF, WAV or FLAC sound differently then there's most likely a bug in the software but could be user error, configureation, etc.  The sound cards only support PCM data, which at that point the data should match the CD.
Quote
2) File size between the two formats is slightly larger with FLAC versus WAV. Anyone know why?
FLAC supports file tagging, which will make the files slightly larger.
Quote
3) "Loss of metadata" and "WAV lack of metadata support" is what consistently appears in print that would seem to give the nod to using FLAC encoding. With dbpoweramp used to rip, and if I ripped to WAV, at what point or how is there "loss of metadata"?
Technically there's no loss of metadata because it never existed on the CD to begin with.  All ripping software looks up the "metadata" from a database to populate the artist, title, song names, etc.
Quote
4) With the BDP-1, is there any practical advantage to FLAC versus WAV using the various control interfaces: mpod, mpad, gnome, mini, max as far as how album art, song titles, etc. will display? I have only used mpod at this point.
The answer will vary by MPD client as some clients don't even look at tags in the file which provides no advantage for FLAC or AIFF tagging ability.    If you use the fat32 file system, there will be some special characters omitted or converted to an underscore "_"  and some situations that will covert short names to all lower case.   i.e. U2 will appear as u2 and Keb' Mo' will appear as Keb_ Mo_.    So if you use WAV (or other tagless file) clients that can read tags will revert back to the directory structure.   So AIFF or FLAC:Lossless would be the better choice IMO.   

I do like the idea of Flac's ability to checksum the data, but I have enough devices where FLAC isn't supported it's not my preference for 16/44.1 files.   For hi-rez files, flac seems to be the best option for now at least. 


Music Matters

  • Jr. Member
  • Posts: 20
Re: FLAC Lossless Uncompressed versus WAV
« Reply #2 on: 5 Feb 2012, 10:20 pm »
Thanks, skunark, for your reply. Doing critical listening, my ears and my wife's ears both come to the conclusion that WAV sounds more "right" than FLAC lossless uncompressed. We have done several listening sessions and each time have come to the same conclusion. Differences are subtle, but I would describe being able to hear into the music a bit more and bass a bit more punchy with WAV. Not a head over heels difference, but to our ears we'd give the nod to WAV. Perhaps using dbpoweramp the metadata issue is not something we need to worry about with either format, as you say dbpoweramp goes out to fetch metadata anyway.

I am assuming that cover art should display exactly the same regardless whether I rip to FLAC or WAV since there is always a cover art file for each album using dbpoweramp. Would that be your understanding, as well? I am also assuming album titles, artist names, and song titles will display the same regardless of which format is used?

Given your experience, is there any downside you see to ripping CDs as WAV?

Crimson

Re: FLAC Lossless Uncompressed versus WAV
« Reply #3 on: 5 Feb 2012, 11:42 pm »
WAV is not portable with respect to its metadata. FLAC saves the metadata as part of its file, whereas WAV does not (it's stored separately). This can be a problem when switching playback programs, or changing machines (a major PITA).

Have you tried AIFF? It's a straight PCM file, like WAV, but can also store metadata as part of the file.

Music Matters

  • Jr. Member
  • Posts: 20
Re: FLAC Lossless Uncompressed versus WAV
« Reply #4 on: 6 Feb 2012, 02:09 am »
Crimson, thank you. I am not familiar with AIFF. As far as sonics, will it be identical to WAV? How does it differ from FLAC lossless uncompressed?

What I am also struggling with understanding is your statement "WAV is not portable with respect to its metadata". I am learning this stuff, and so far with the BDP-1 I am using my iPhone with MPoD to control - just very basic. I am planning to get an iPad and then use MPaD. What practical differences will I experience with album art, song titles, artist names, etc. as far as AIFF versus WAV? Is this the metadata you are referring to that is not portable? Thanks for enlightening my understanding.

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #5 on: 6 Feb 2012, 08:02 am »
AIFF and WAV are both uncompressed PCM containers.

WAV doesn't have a standard way to store the metadata so each application stores the data differently and that metadata doesn't normally transfer.     

Apple Lossless(ALAC) another popular format, is more like FLAC where it can compress the lossless stream to save disk space.    Really the only advantage one has over any other for 16/44.1 files is usability.    I don't recommend ALAC with the BDP because certain files I've had issues with and since migrated to AIFF for CDs I've ripped.   

Using NIN The Slip downloads I don't hear a difference between the 24/96 FLAC and WAV files nor with the AIFF (converted from the WAV file).   I do feel the bass is sharper when compared to the CD or even the 16/44.1 AIFF file that I ripped.   With the BDP I find that all files sound better than other players and have pretty much stopped with that for now.    For kicks I have mixed both high res and cd lossless files of varying formats on the BDP and have yet to reliably pick them.  It's easier to spot hirez vs cd lossless but I'm still not perfect spotting them, but that is causal listening not a critical comparison. 

zolta

  • Jr. Member
  • Posts: 2
Re: FLAC Lossless Uncompressed versus WAV
« Reply #6 on: 7 Feb 2012, 01:34 am »
i have seen this topic on many forms, and I am always surprised that someone would hear the difference between the two file types.  especially in this case where the flac file is not being compressed.  If a cd is ripped with the same hardware and software and then using the same hardware and software to convert the file to a wav file and then to flac, such as how EAC does this task, there should not be a difference in sound quality as it is all one's and zero's.  I have never been able to tell the difference between slightly compressed flac and wav.  There was a time when i was keeping both, so my portable could play the flac as they were smaller in file size, when i up[graded my DAC i did comparisons and could not hear the difference, and got rid of the wav files, to save space.  I would be curious to know if the the wav file sounds as good or better than the cd, which is my case.

dlach

  • Jr. Member
  • Posts: 9
Re: FLAC Lossless Uncompressed versus WAV
« Reply #7 on: 16 Feb 2012, 04:11 am »
With their March, 2012 issue, The Absolute Sound magazine concluded a 4-part project to evaluate and quantify every sonic variable in computer audio.  This includes comparing sonic differences between FLAC and WAV files, different ripping s/w, playback s/w, etc.  The authors, Dr. Charles Zeilig and Jay Clawson, offer specific recommendations that may surprise most of the folks on this thread.  For example, they claim that CD's ripped to WAV do in fact sound better than the same CD ripped to FLAC.

I'm going to try some of their recommendations and decide for myself.  The March, 2012 issue should be available on newsstands now.

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #8 on: 16 Feb 2012, 07:34 am »
With their March, 2012 issue, The Absolute Sound magazine concluded a 4-part project to evaluate and quantify every sonic variable in computer audio.  This includes comparing sonic differences between FLAC and WAV files, different ripping s/w, playback s/w, etc.  The authors, Dr. Charles Zeilig and Jay Clawson, offer specific recommendations that may surprise most of the folks on this thread.  For example, they claim that CD's ripped to WAV do in fact sound better than the same CD ripped to FLAC.

I'm going to try some of their recommendations and decide for myself.  The March, 2012 issue should be available on newsstands now.

Just means the FLAC codec isn't as mature as the WAV then.   They all have the end-goal of producing a bit-perfect PCM stream.   It's real simple to check to see if the conversion is bit-perfect and also real simple to test the jitter at the output, for a given system.  If they don't provide this then their findings is just more hearsay, chances are they won't provide this detail since it's not even financially beneficial for them.   I will have to go to the newsstand though and read that article.

NMG

  • Jr. Member
  • Posts: 60
Re: FLAC Lossless Uncompressed versus WAV
« Reply #9 on: 16 Feb 2012, 10:17 pm »
If you go on Audio Asylum - Digital and Computer Audio, several engineers who would know say that the articles are full hooey. I have never been able to hear any difference between a CD ripped with DBPA into FLAC or Apple Lossless.

Neal

saisunil

Re: FLAC Lossless Uncompressed versus WAV
« Reply #10 on: 16 Feb 2012, 11:18 pm »
There is so much to do with Transport as well ... PC vs. Mac OS to drivers to hardware to optimization

and who knows what ...
and then resolution of DACs and Players ...
and then some people are able to and care for the subtle differences more than others ...

There is so much more to computer audio than we / I currently know ... there is so much to explore ... if some people hear those differences - great - perhaps we can try or not care about it ...

Does not mean those who do hear the difference are lying or have a screw missing upstairs ...

Heck there are people who believe bits in and bits out are the same - whether they come from ipod or from high end transport ...

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #11 on: 17 Feb 2012, 12:03 am »
There very well could be differences, but a better comparison would be PCM vs DSD.  PCM is contained in the  WAV, AIFF, FLAC, ALAC and other files, different codecs, drivers and hardware can and probably do have bugs.   

So get a known good bit-perfect PCM stream and compare that to a known-good bit-perfect DSD stream.  That would be an interesting debate.     

Even checking the quality of various codecs for a given file type would also be interesting.  A good example is the nero AAC codec, vs Apples vs the less mature opensource AAC codecs.   Of course, folks wouldn't provide quality or performance benchmarks which is the key information and is much like you see with video codecs for TV and blu-ray reviews,

Music Matters

  • Jr. Member
  • Posts: 20
Re: FLAC Lossless Uncompressed versus WAV
« Reply #12 on: 10 Mar 2012, 12:44 am »
Since I started this thread, I thought I'd report back. I abandoned all of my FLAC Uncompressed files and reripped using dbpoweramp to WAV and AIFF. I archived the WAV files and loaded the AIFF on the SSD external drive tethered to the BDP-1. I agree with the conclusion reached in the recent series of Absolute Sound articles comparing WAV to FLAC - that WAV exceeds the audio quality of uncompressed FLAC by a margin of 10% or so. It is particularly noticible in the lower end - overall pace of the music is improved with WAV or AIFF and the soundstage similarly improved by a margin. So I thank skunark (above) for the suggestion to use AIFF and for anyone starting out, I strongly recommend AIFF or WAV. I chose AIFF because of its better metadata support capability. Unless there is a compelling reason for you to chose FLAC, I wouldn't go that route. Have fun!!

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #13 on: 10 Mar 2012, 07:41 am »
Interesting findings. 

I know the BDP will work harder with FLAC and ALAC files since it has to uncompress them before shipping the lossless PCM data to the sound card.  You can watch the processor utilization and server load and see how much extra effort it takes.  That extra effort consumes more watts which is translated to heat, it's a pretty minute difference but maybe that plays a part.  In the end FLAC, ALAC, WAV and AIFF should be bit-perfect when sending the PCM stream to the sound card, so more likely there's a bug with the FLAC codec and probably something to revisit if there's any significant firmware updates with the BDP.   I haven't tried Flac:uncompressed but I believe it is a recent feature that might require a codec update for the BDP to take full advantage of the tweaked format.

rbbert

Re: FLAC Lossless Uncompressed versus WAV
« Reply #14 on: 10 Mar 2012, 02:22 pm »
With their March, 2012 issue, The Absolute Sound magazine concluded a 4-part project to evaluate and quantify every sonic variable in computer audio.  This includes comparing sonic differences between FLAC and WAV files, different ripping s/w, playback s/w, etc.  The authors, Dr. Charles Zeilig and Jay Clawson, offer specific recommendations that may surprise most of the folks on this thread.  For example, they claim that CD's ripped to WAV do in fact sound better than the same CD ripped to FLAC.

I'm going to try some of their recommendations and decide for myself.  The March, 2012 issue should be available on newsstands now.

As has been fairly well chronicled in many places on the Internet, including at least two other AC threads, the biggest problem with these recommendations is that the WAV > FLAC > WAV conversion process itself causes sonic degradation, an easily disprovable hypothesis.  So while real-time FLAC decoding and playback may (or may not, depending on hardware and software) be associated with altered sound quality, the magazine's articles in no way establish that.  Although I don't have a BDP-1, I 'd be a little surprised if a dedicated music player like it has audible problems decoding FLAC, but I don't really know about that.  If it does, a firmware update should improve or even fix it.

groovybassist

  • Full Member
  • Posts: 505
Re: FLAC Lossless Uncompressed versus WAV
« Reply #15 on: 10 Mar 2012, 03:16 pm »
I had previously ripped my CDs to FLAC and decided to do an experiment a couple of weeks ago - I ripped one well recorded cd to WAV, keeping the FLAC files. I then sorted the tracks, so that they alternated FLAC, WAV for each tune. While a casual listener wouldn't care about the difference, critical listeners definitely will. The most obvious thing was the soundstage collapsing to 1/2 size on the FLAC files. The WAV files had a much greater feeling of being at the performance. Subtle cues likes snares on a drum and differences between cymbals were much more obvious on WAV. I didn't want to reach this conclusion, as it meant I needed to re-rip all my stuff, but I'm in the process of switching everything to WAV.

rbbert

Re: FLAC Lossless Uncompressed versus WAV
« Reply #16 on: 10 Mar 2012, 08:03 pm »
Rather than re-ripping, did you try just decoding your FLAC to WAV?  I haven't tried ripping directly to FLAC, but I have ripped to WAV, then compressed to FLAC for storage.  Uncompressing the FLAC back to WAV then results in a WAV file identical to the original WAV both sonically and by data checksum.

Phil

Re: FLAC Lossless Uncompressed versus WAV
« Reply #17 on: 10 Mar 2012, 09:28 pm »
interesting results.  have to give this comparison a try with the BDP-1.  Did a comparison with AIFF, WAV and FLAC when I was using a Touch and preferred them in that order, but the BDP-1 could be different. 

FLAC has been sounding very good on the BDP-1...

Phil


calinet6

  • Jr. Member
  • Posts: 2
Re: FLAC Lossless Uncompressed versus WAV
« Reply #18 on: 10 Mar 2012, 09:30 pm »
Hi. Computer scientist and algorithms expert (and audiophile) here to chime in. Interesting discussion so far, but I thought I'd respond to a few points. To preface this, let me say I am all for improving audio and the way we listen to it, and I find differences in DACs and transports and digital processing solutions to be a very worthwhile pursuit and I'm endlessly fascinated by those kinds of optimizations. However, FLAC vs. WAV is not one of them.

Quote
Just means the FLAC codec isn't as mature as the WAV then. They all have the end-goal of producing a bit-perfect PCM stream.

They don't just have the end-goal of producing a bit-perfect PCM stream, they both do produce a bit-perfect PCM stream. There is no problem with the FLAC lossless codec or its implementation. I know, I've tested it, and I've verified the algorithm and the underlying code. The digital stream coming from a FLAC file isn't just supposed to be identical to the WAV, it is identical, and provably so.

Quote
WAV exceeds the audio quality of uncompressed FLAC by a margin of 10% or so.

This is scientifically impossible. Were there a failure in the FLAC decoder or in the FLAC codec, the bitstream going into the DAC would differ. It does not, ever, 0% of the time, once again, provably beyond a shadow of a doubt.

Quote
... several engineers who would know say that the articles are full hooey.

This is 100% correct. The articles are describing a difference which is physically impossible and can only be the result of false perception or conflicting variables in the testing process.

Quote
While a casual listener wouldn't care about the difference, critical listeners definitely will. The most obvious thing was the soundstage collapsing to 1/2 size on the FLAC files.

This, too, is physically impossible. The digital stream output by your DAP is going to be 100% exactly the same from a FLAC or a WAV file, regardless of the decoding process or the format. They are the same bits, and bits are either on or off—there is no in-between. It's not possible. The only limiting factor in the digital processing is the transport and the DAC, which both come after the PCM stream is generated, and the PCM stream generated by the WAV and FLAC files from an identical source are mathematically provable to be identical. In short: if you hear a difference, you are making it up.

I realize this may be difficult to understand when we're used to describing such small and subtle differences, many of which I fully support, but this is one case where having a little knowledge of algorithms, mathematics, and provability comes in handy. This basically means that in the realm of computers and digital processing, there are certain things you can't argue against, things which are mathematical facts. And the provability of identical bitstreams is something we understand so very well that we can say that this is a mathematical fact, no ifs ands or buts about it.

Please let me know if anyone has any questions about this, I'd be happy to help you understand further. In the meantime, please feel free to encode all of your WAVs or CDs in FLAC format, and as long as you have players which support it, rest assured that your bit-perfect audio will remain at 100% of its original quality.

skunark

  • Full Member
  • Posts: 1280
Re: FLAC Lossless Uncompressed versus WAV
« Reply #19 on: 11 Mar 2012, 09:56 am »
They don't just have the end-goal of producing a bit-perfect PCM stream, they both do produce a bit-perfect PCM stream. There is no problem with the FLAC lossless codec or its implementation. I know, I've tested it, and I've verified the algorithm and the underlying code. The digital stream coming from a FLAC file isn't just supposed to be identical to the WAV, it is identical, and provably so.
Experienced EE here :)

Is there such a thing as bug free software?  I'm rather shocked anyone would consider flac or any codec bug free.   I assume your findings with flac is limited to a small subset of tests, but please do post your results, would be interesting to see.  I assume you are not aware about Flac's changelog and their bug tracker that can disprove your outcome.  There's several bugs there and that's not even the complete stack, so who knows what other bugs exist upstream or downstream of the codec!   Every codec is bound to have bugs even if the algorithms, mathematics and provability are sound...  Even how the software is compiled can introduce bugs which is why you see a big push for integer only compiles for these codecs.

I have experience issues with ALAC files on the BDP-1 that doesn't exist with other players.  These issues include hangs partway through a track and sometimes would pops/glitches between tracks and all reproducible.  Clearly some issue with ALAC and the BDP.  You can google about the quality of various AAC codecs and the issues there, but that is lossy format and will never be bit-perfect so might be a little uninteresting.   

A broader set of regressions with various hardware and software combinations will indicate there are bugs....

Quote
Quote
WAV exceeds the audio quality of uncompressed FLAC by a margin of 10% or so.
This is scientifically impossible. Were there a failure in the FLAC decoder or in the FLAC codec, the bitstream going into the DAC would differ. It does not, ever, 0% of the time, once again, provably beyond a shadow of a doubt.
Sure in academia where everything is ideal, you can formally prove they are equivalent, but in the real-world, there's a few more twists and turns.  I agree WAV, AIFF, ALAC and FLAC should be bit-perfect, but a difference might indicate issues, and yes it very well could be PEBKAC, but could otherwise indicate an actual issue for a given scenario.  My first hand issues with the ALAC files on the BDP does make me question all formats and all devices now.  I even greatly questioned FLAC for a spell since I had ALAC and FLAC intermixed.  I don't think any less of ALAC, and don't even have an issue using it on other devices, I just won't use it with the BDP until firmware is updated that has an upgraded codec.    Even Squeezebox users have reported issues around various codecs and how everything gets converted to FLAC before streaming.   

There are a few manufactures that will provide a test file or test cd that is used in conjuction with the DAC to test the quality of the transport.  (I believe you can do this with an HDCD if your DAC indicates it)  That's pretty nifty test, it's not a complete test by any means, but useful nonetheless. Out of the several thousand ALAC files I had, very few caused the hang or the glitches between tracks, we are probably talking less than .2% of a hard failure-rate that would easily fail any bit-perfect test for a transport.  Instead of debugging this further with ALAC, I just opted to switch to an uncompressed and more mature format.  Using the notion that all lossless formats are bit-perfect, assuming no bugs, I did switch to AIFF since it supports tagging.   

Now why would one feel that one lossless format is X% better than another, vs my experience where there was a huge failure in playback?    I think the squeezebox users probably could shed some light on this too.   And the OP is comparing flac:uncompressed to wav and aiff, so that's even yet a closer file format.   Perhaps it has to do with features like volume normalization with the playback (BDP is suppose to have that off with MPD) that is only enabled for one codec?  Perhaps there is a higher CPU load with one format over another burning more watts that could be raising the noise level.... tons of little possibilities...   I would agree that it's highly unlike there is an actual difference between the uncompress lossless formats, but I also don't mind scratching my head pondering why it could be different. 
« Last Edit: 11 Mar 2012, 11:18 am by skunark »