FLAC vs WAV


I have observed (heard and then tested so as to confirm) the following “condition” as it relates to the widely debated issue of FLAC quality. The purpose of this topic is to gather opinions as to whether or not your observations are similar too – and therefore support – my own.

It is widely understood and accepted that a FLAC file while “compressed” is “lossless” as compared to its corresponding WAV file. Let’s assume (i.e. not debate) this is completely true. What I am noticing is that when the FLAC file is “played” via any FLAC player it sounds different from the sound of the “same” (equivalent decompressed FLAC) WAV file when played back via the same player that was used to play the FLAC file. This is specifically noticeable (to me) in the low frequency spectrum. The WAV has considerably more “sonic energy” that manifests itself as appearing to be a bit louder, wider in frequency range and perhaps even dynamic range as compared to the FLAC equivalent.

I’m curious as to your findings when you compare a FLAC file played natively as compared to the WAV equivalent played via the same player (for example, play both the FLAC and WAV via VLC media player) or practical equivalent, such as if the FLAC was burned to CD and you are comparing the FLAC played via VLC and the CD played via a CD player.

I am further assuming that the WAV file is a more accurate representation of the audio than the FLAC. This is to say that should you agree with the aforementioned, it would be preferable to play the WAV file or decompress the FLAC file before using it.

128x128gdhal
Lossless is lossless. Format should not matter.  If it does then gear is doing a better job with one than the other foe some reason.   I have used both and found format has not mattered.  
I too have noticed for a long time (first with PS audio PWD1/2 and now with thier DS) that flac does not sound as good as WAV or AIF. This was running first with Elyric and now Jriver on an old Mac pro and now a new Mac pro. Transcoding to wav on Elyric on the fly sounded better than streaming flac, but pre converting the flac to AIF file sounded the best. When I buy HD tracks now, I choose AIF.
none of this makes sense, except possibly the extra CPU overhead degrades the sound. Hard to believe with how much power my 6 core Mac pro has, but then again many things in this wacky hobby seem counter intuitive. (like the improvement I got with cat 6a ethernet replacing cat 5).
I am glad to read someone else has noticed this.  In thinking about it I concluded that there are still a lot of variables between the formats.  The codecs, cpu time, players, throughput.

I have settled on AIFF for now but I hope to learn more by reading other's experience and practices.


Thank you mapman and emailists. I concur with you Emailists, although I have not used AIF with a MAC. Perhaps to restate my point of this thread in different words, what I am saying is that it certainly appears as though it is "best" to pre-decompress the FLAC first - using say Traders Little Helper - and then once decompressed to WAV then play it. Perhaps FLAC players which decompress on-the-fly while playing does not do as good of a job as pre-decompressing. So mapman, while I agree that lossless is lossless, in my opinion (and emailists and perhaps others) while the file is lossless it may not be as good as when decompressed into its then uncompressed lossless state. 
It has to be converted to standard PCM format which is was format before it can be converted to analog.  So there is an extra step in the process.   If done properly it should not matter but may not be in some cases.  But the information content of the source is the same assuming the software that encodes the flac does its job correctly.  

So Flac may not sound as good in some cases or vice versa but if so its not because of the format.   Just like how different cartridges may sound better or worse playing the same record.  
Computer decompresses data, places it in the memory and adds timing to create and output S/Pdif (PCM). The only difference, assuming bit perfect conversion, can be jittery S/Pdif timing, that would add noise to music but dynamic range should be the same. Since computer timing errors would be very small noise would be added to mid and high frequencies. I cannot imagine how it could affect the bass.
Just as you assume flac is indeed lossless, I think it is fair to assume that the decompression algorithm on any reasonable player does its job correctly.

Have you tried zero compression flac? It is not compressed but has a flac header?

Systems that play from memory can decompress a flac file and then put the PCM into memory for playing. At that point, the computing to decompress is not important. Have you tried that approach?
DTC, the short answer to your questions is no.

The longer answer is as follows. The FLAC file(s) I am speaking of are "created" outside of my control. Specifically, I download them from bt.etreeg.org using Utorrent (64-bit). I then burn to CD using Ashampoo version 6. Playback via Emotiva ERC-3, Musical Fidelity M6si, Golden Ear Triton One's. It as at that point in some instances (music passages) I can notice more "bass energy" (simplistic term but that will suffice for now). Then, I have compared that point (time and track data) in order to play back the FLAC and decompressed WAV version of it via my computer, Gateway NV79. That connects to my M6si via USB. Granted, the DAC on the M6si differs from the DAC in the ERC-3, however, that should have nothing to do with this because what I am finding is that the WAV when played back is more representative (all around but looking at the bass energy here) to the ERC-3 playback than the FLAC. This is to say the FLAC and WAV when played via the same computer and at the same track and time index produce different (in some passages that I can detect using my ears only) bass energy. Also, I would be remiss if I didn't point out that I use VLC player (latest version) and the "settings" are kept the same when playing.
The comparison between using the ERC-3 or the PC as a transport  is, of course, not a good way to make the comparison. It sounds like you also compare the FLAC and WAV both played back from the same computer and using the same equipment. That is a valid comparison. If you are doing that, then you should be able to convert the flac to flac with zero compression as a test. You might also want to try a couple of different players, to verify that it is not some issue with the player.

This issue has been discussed hundreds of times. Some people say they hear a difference, most say they do not. If you near a difference, it is more likely to be an issue with your player or the PC than with just the format. That is one reason I suggested zero compression. It might also be interesting to try different compression levels, like 1,2,3 up to whatever you are currently using.
You just use a conversion program or ripper that outputs a flac file. When you output flac there should be a compression option, with 0 as one of the options. The default is usually level 6. I would use JRiver, the player I use. I am pretty sure dBpoweramp does it also. Whenever you create a flac, you should have a 0 compression option.

Apart from this discussion, it is a nice way to use wav files but maintain the meta data capabilities of flac.
EZ CD Audio Converter has 0-Fastest, 5-Default, and 8-Smallest
dB Poweramp has Lossless Uncompressed, Lossless Level 0-Fast, and 8-Highest. Which one do I use?

IF you have a resolving system (most think they do, but they don’t), you will hear this difference, PROVIDING that you are using a playback software such as Foobar, Jriver or Amarra on a PC or Mac and outputting via USB or S/PDIF.

We can argue this until the cows come home, but its simply true. I don’t know why. I can speculate, but I have no proof.

So, how can this problem be eliminated?

Possibly by using a network server that eliminates the use of the audio stack in the computer or server.

Possibly by purchasing a well-designed server like the Antipodes from New Zealand. I did this comparison at a trade show and it is the ONLY server I have used where the FLAC and WAV files sounded alike.

I am not affiliated with the guys making Antipodes servers.

BTW, I have made this comparison with AIFF, ALAC, FLAC, uncompressed FLAC and they all sound compromised compared to WAV.

Steve N.

Empirical Audio

Steve (audioengr), given your response, would you go further to say that by extrapolation, one should convert FLAC to WAV prior to burning to CD (as opposed to letting the same CD burning software do the conversion)? Again, in my case I would convert the FLAC to WAV using Traders Little Helper and then burn using Ashampoo V6 (Windows 10). This would be opposed to not using Traders Little Helper and allowing Ashampoo to burn the FLACS to CD (as audio CDA files) directly.
NAIM Uniti rips to WAV only..maybe that is WHy they made that decision ( long ago)... they do seem to place a priority when voicing on bass drive.

One thing you may want to do is precise level match check with your FLAC to WAV comparisons including ripping some tones, the Stereophile test disc and or a tone generator might be a good starting point....as a thought.

cool thread and civil discussion..appreciated..

I personally would feel more comfortable converting FLAC to WAV and then use that file to burn a CDROM.  I would recommend DBpoweramp. I dont use CD's anymore myself because I can reduce the jitter more by using computer playback.  If you rip to CDROM, I would recommend Mitsui Master Gold disk.


Steve N.

Empirical Audio

From the dBpoweramp site

"Compression affects how much effort goes into compressing the audio, all compression modes give the same decoded audio (it is lossless after all), the higher compression levels will give a small % file size saving, but will require more time to compress and decompress. Compression Level 0 requires the least compression time, whilst Compression Level 8 the most. Uncompressed is a special compression mode with stores 16 bit audio in an uncompressed state."

So, uncompressed sounds like the best for 16 bit data, although I have never tried it.

No experience with EZ CD Audio Converter.
Computer should decompress files before putting them into memory buffer. Up to this point it is just a data, that can be verified by the checksum. Music starts when timing is added. In case of S/Pdif it is done in computer but in case of asynchronous USB it is done on the receiving end (DAC). Computer or data format cannot affect it unless big electrical noise can flip-up the bits that are being send (not likely).
One should becsurevwav sounds better in their specific setup before committing.  Flac played right should have no difference.  Use hardware and software designed for high quality flac streaming and you should be fine    I used wav for a few years then decided flac was the way to go and converted everything.  Both sound similarly excellent imho in my setup.  Flac supports flexible tagging which enhances user experience. Wav does not.   
No difference unless your DAC can't handle input jitter. A suprising number of snake/ I mean audiophile DAC can't  handle input jitter properly.
The diferrence in sound between FLAC and WAV/AIFF is real. I hear it too.
People confuse lossless as the only contributor to sound quality for digital files. Losslees only means that the file is bit-perfect i.e. No loss of digital bits in the file. But another important factor is whether the file itself is compressed or uncompressed.
But the difference in sound quality that you hear between FLAC and WAV/AIFF is not because it is lossless or lossy. It's because FLAC is compressed while WAV and AIFF is uncompressed. On playback of FLAC files, the computer needs to uncompress the files and this extra processing can add noise and phase distortions. 
With some CD ripping software like dBPower, you can set the compression level of FLAC between 0 to 10 with 5 being the standard compression level. If you rip unto FLAC with 0 compression level, it becomes uncompressed and lossless. And it sounds similar to WAV and AIFF. Try it yourself.
If you can hear difference between flac and wav you need to get a properly designed DAC, eg, one with a reclocker built into it like the Emotiva DC1. Lossless is lossless, there is no difference between decoded flac and wave output other than jitter values (flac will have more jitter because windows is not a real time operating system)

coli - this has nothing to do with jitter, besides, there are no DACs that are immune to jitter, even those that reclock.

Steve N.

Empirical Audio

I'm listening to FLAC right now.  Everyone's digital should sound so good.
I have found little or no difference between the 2 formats
Much has to do with the process I use DB power amp for trsnsfer at highest uncompressed level 8. The dac also has a say in how resolved if is . The Digital cable even more 
So. Guys if you have not tried the Curious USB csble you have not heard your music.
This cable is very musical not flat like the standard even well respected cables 
Analog finally has arrived for digital.
The USB regen, as well with Curious on any usb cable down stream .
Unless you have experienced this ,then you have No educated way to make a comment. I would not recommend this if not a statement product.
Better then both Wireworld platinum,or AQ Diamond and much less monies !!
audioman58 - Level 8 is the most compressed version of flac, not the highest uncompressed level.

This is a bigger issue than just cables and  regenerating usb signals.  You may want to do some reading on these issues.
My Wireworld Platinum Starlight USB cable is way better than the Curious cable. The Curious is too foward and bright in the mids. The Wireworld is much more relaxed and open with a much bigger soundstage.
On the grand scale of things that make a difference in sound quality, I have found digital cables and file format (.wav versus flac) to be pretty much the least significant factors. I'm sure YMMV.
Without going into great details I would say I have an above average system.  To me wav. is better than flac.  Not a lot, but better.
ddriveman made the perfect post. For discussion across everyone's systems his points reach the conclusions that will usually apply. Bottom line use AIFF or WAV, if if metadata headaches bug you, like they do me, use AIFF. Cheers,
Spencer
One more vote to WAV files, altough I found sacd and high fidelty pure audio( blu ray) better in that order, so even most of my music are on cd, i prefer sacd.
I have been looking at flac compression levels some more .Let me clarify uncompressed flac, since I confused flac 0 and uncompressed above.

Flac 0 is not uncompressed. It is the lowest level of flac compression, but it is still far from uncompressed. Many packages create flac 0, but dBpoweramp is the only one I know of that does truly uncompressed flac. There may be others, but dBpoweramp is the one I know.

As an example, I have a track that is 69KB (39%) at flac 6. It is 73 KB (42%) at flac 0 and 122KB with uncompressed flac. AIFF and WAV are also 122KB. So, if you want flac with no compression you need uncompressed flac, not flac 0.

Flac and WAV both have well defined meta-data capabilities. WAV originally did not include meta-data, but many implementations do include meta-data today, although the implementations can vary. So, if using meta data, flac and AIFF are the preferred formats.

Conclusion - If you want an uncompressed format use uncompressed flac (not flac 0), or AIFF, with WAV an option if you understanding the tagging issues.


The higher the compression level, the more compute time  needed to do the compression. However, people who understand the flac compression algorithms say that the time to decompress the various levels (0 to 8) are basically the same. That may not seem intuitive but there are good explanations from the for the people who know. Uncompressed flac, however, should take not time to decompress, since it is not compressed.

Hope that helps clear up flac 0 and flac uncompressed. They are different.


As a relative newbie in digital world, I recently elected to rip my cd collection to have access to everything without going through hundreds of cd's. Using dB I started ripping at first in the flac default 5 compression. Playing back through my Magnum Dynalab MD 807T, I must admit that it sounded pretty good (at the time I did no A/B tests) After ripping many and coming to see that I did not need the extra compression due to using a external 4 drive enclosure for storing files, I deleted everything and re-ripped in flac uncompressed.

Having access to everything is a pleasure, but accessing through the computer seemed to be occasionally problematic. Errors popped up here and there (maybe once for every four hours of streaming). It presented itself as what I would describe as dropped bits, momentary silence etc. I then elected to move everything to a NAS and take the computer out of the loop for anything but operation control through JRiver. So I access the library through desktop/laptop or tablet and these momentary errors have evaporated.

Always one to read, experiment and verify, I am going to rip some cd's that are already in flac uncompressed format and sound exactly like the original material into WAV and playback using the same method above and see if there is any  difference that I can detect in my system. If not, I will consider myself lucky and continue using flac uncompressed and just get larger drives, if needed down the road.  

As a side note, I have been experimenting with changing playback parameters in the PS DirectStream, since it was recently added, to find optimum setting, there also there is going to be a learning curve but that's what a hobby is all about 
PS DirectStream is the one DAC that doesn’t handle input jitter at all, you will hear differences all the time with that DAC. Even $200 consumer receiver handles jitter better than the PS Audio DAC.

I’m so glad I was able to get rid of it at a big loss to me.

btw: jplay reduces jitter of PC audio output, you’ll hear a difference with that DAC. On the other hand with a properly designed DAC like the Emotiva DC-1 which has a built in reclocking circuitry, there will be no differences what so ever between formats/players.
DTC,

You are correct about FLAC compression level 0 and uncompressed FLAC. FLAC 0 is not uncompressed. I stand corrected. And dBPoweramp offers uncompressed FLAC.
ddruveman - I think I started the idea that  flac 0 is uncompressed here. But now we know that flac 0 is not uncompressed. I have have had it wrong for a while. You need to go to dBpoweramp to actually get uncompressed. Personally I still used flac 5 or 6, since I cannot hear any difference.
jwm - look at my data above. Flac 0 is less compressed, but not by much - 39% for level  0, 42% for level 6 for my file. And, according to the flac gurus decompressing for 0 or from 5 or 6 takes essentially the same amount of processor time.
And given that there is division between FLAC and WAV among certain of those who have responded to this topic, I’d now like to ask if it is the opinion of those in the FLAC or WAV camp as to whether or not there is a quality difference in playing and/or decompressing FLAC depending on the original compression level.

The reason I'm asking is not simply to fuel yet another debate. In my case, I obtain the flac files from bt.etree.org. I (seemingly) have no visibility into the compression level. In fact I have never read a notation from any of the posters on that site who distribute in flac (both 16 and 24 bit) having anything to do with the compression.

What I have found is that when I download a show in totality it is (hypothetically) 600 meg and when I convert/decompress to WAV it is closer to a 1.5 gig.

Thanks to everyone.
I believe, Reference Recordings primarily produces HDCD discs. I would not be surprised if he used wav in his processing. He would have little reason to compress data while doing the engineering. Can you point to a reference about Johnson preferring wav? It would be interesting reading.
Steve N., you have previously stated that to you WAV sounds better than AIFF.  Do you still feel that way and/or have any idea why?
I use XLD for ripping.  Playback is via an Ayon S-5 pulling files from a NAS drive.  Playback control is via JRiver.  Before ripping my complete CD library I tested the AIFF and WAV formats of some of my usual demo CD's.  AIFF gave superior performance - significantly more separation of voices and instruments in the soundstage and faster transients, especially in the attack on acoustic strings.  From my experiment I concluded which format sounded better on my system but the only recommendation I would ultimately make would be to test the various formats of interest on one's on system for as the saying goes, "Your mileage may vary."
Something interesting (to me anyway) I have discovered. As reported by Windows Properties size (not size on disk), a WAV encoded to Flac (compression level 6 but realistically any) and then back to WAV from that FLAC is not identical. For example, if the WAV is 399,718,326 bytes and converted to FLAC it is 188,443,075 bytes. If I then decode the FLAC back to WAV the size is 399,718,324 bytes. Can that 2 byte difference be explained?

This will be debated as long as digital is in existence.  I have limitless storage capability (10TB) and at the time, all I had was time (was in-between jobs).  I spent endless hours testing CD rips to see which was better before I ripped my entire library.  I ripped dozens of CDs in WAV and FLAC, and also AIFF.  I used dBPoweramp to rip.  As much as I wanted to hear it, I could hear ZERO difference between them all.  Zero, none, nada, zilch, nothing!

If someone can hear the difference, bless you.  My system is incredibly resolving so that's not it.  Just my experience.

What I did hear a huge difference in was the playback software.  JRiver was best, Mediamonkey (?) was horrible, iTunes and a whole bunch of others were tested.

acurus, thank you, however, what I have discovered transcends whether or not you, I or anyone else can "hear" a difference. Question is, "is there a difference". Apparently there is.

By the way, I’ve read great things about JRiver from numerous audiophiles so do not doubt it is very good. Best is subjective. In my case I use Windows Media 12 (on Windows 10) and find it more than adequate.

The difference of 2 bytes is probably just a matter of how each program pads the end of a file with zeros. It depends on the way the program writes the data. It has nothing to do with the audio data. It is just a programming difference.
You can do a binary compare of your two files using, for example,  the Windows fc command at the command prompt.

fc/b  file1 file2

There are other compare options, but this is a place to start if you want to go into the differences in your files in more detail.