Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[compcond]

Does it make any sense that .wav files sound significantly better than .flac?  I keep telling myself I'm crazy but I keep hearing a significant difference.  Same CD... ripped twice in succession.

Rip .flac at quality 6.  MC18 set to use WASAPI - Event with all enhancements turned off.  Benchmark DAC1 HDR, Conrad-Johnson MF2500a, custom built speakers DynAudio D210, D52AF, 2x17WLQ, HSU VTF3-MK3.  PC is Intel DQ67, i5-2400, W7-64, 8GB RAM.  USB interface.

[DarkPenguin]

No.

[dean70]

Yes noticed the same thing myself (execpt with aiff format). There is a slight improvement in sound quality. Even though they are both lossless formats, I believe it is related to software induced jitter - the souce bitrate jumps all over place with flac files, but is constant with aiff or wav.

[contium]

Doesn't make sense but I have heard more than a few people (well, maybe just a few) over the years make similar claims. I keep meaning to test it myself but have yet to get around to it.

[DarkPenguin]

*chop* ... the souce bitrate jumps all over place with flac files, but is constant with aiff or wav.

That's because the one is a variable bit rate and the other is constant.  Or, another way, one is compressed and the other isn't.

Edit: I should note that I'm only commenting on the bitrate jumping around and not any audio differences you might imagine hearing.

[jmone]

I wish we had an ABX plugin for MC ....

[Matt]

If you use MC, they will sound identical.

If you hear a difference, think critically about your test procedure.  In particular, make sure you don't know which format is playing when you evaluate.

[jmone]

If you hear a difference, think critically about your test procedure.  In particular, make sure you don't know which format is playing when you evaluate.

Like an ABX Plugin? 

Matt (representing the logical group) is right that it should sound identical....but lots of people claim they can pick a difference.  The good thing about ABX is that not only don't you know what format or setting is used but it makes it hard to cheat.

[rayooo]

I've been through this and through this now for a couple years, and I've still got a mix of FLAC,  WAV and a couple other formats on the NAS to prove it.  
Other than my on imagination from time to time, I personally can't identify any difference, which must mean I don't know what I'm talking about and/or can't "hear".  
..regarding the old "Trust your ears"... I do, I trust my ears to deceive me. actually it's not my ears, it's the audio perception I have on a given, day, minute, hour, second based on dozens of psychological and other factors like mood, expectations, time of day, suggestions from "trusted advisers" etc.etc.  JMHO, YMMV, Trust your ears...

If you use MC, they will sound identical.

If you hear a difference, think critically about your test procedure.  In particular, make sure you don't know which format is playing when you evaluate.

[6233638]

There's no difference between them. Losslessly compressed, or uncompressed audio, will sound the same.

There are many reasons why you might hear a difference, but it's all in your head. This video goes into detail about some of that.

[rayooo]

Fantastic Video!  thanks!

There's no difference between them. Losslessly compressed, or uncompressed audio, will sound the same.

There are many reasons why you might hear a difference, but it's all in your head. This video goes into detail about some of that.

[Mikkel]

No  . Download Foobar and make ABX comparisons and tell us the result :-). My bet is you cannot hear a difference then.

[kstuart]

No  . Download Foobar and make ABX comparisons and tell us the result :-). My bet is you cannot hear a difference then.

You can't use Foobar's ABX to compare, because it converts to WAVE as part of its procedure.

I have never compared wav and flac...

... but someone did a blind test at conventions, using many different subjects.   What they discovered is that the faster the computer, the less people were able to tell the difference.

What that indicates is that what dean70 said earlier in the thread is correct - it is likely software-induced jitter.

By the way, the idea that "logically this cannot exist" simply means that you think you can keep every hardware and software aspect of the playback process in your head, which is "unlikely".

[dean70]

It might also tie in with the JRMark score - I have a moderate score of 1840. Maybe on a faster machine there would be no difference  ?   

On another note, I tried a certain player whos name I wont mention that a few rave about, but found no difference or benefit in sound quality.

[bobkatz]

It might also tie in with the JRMark score - I have a moderate score of 1840. Maybe on a faster machine there would be no difference  ?   

On another note, I tried a certain player whos name I wont mention that a few rave about, but found no difference or benefit in sound quality.

Well, the first thing to do is to confirm 100% that the copy is identical. I do this by bringing both of them into a reliable DAW, lining them up to the sample, then inverting the polarity of one, and cancelling. If there is 100% (measurable and audible) cancellation throughout the length of the piece, then the copy is identical. At that point (and it's highly likely you are at that point), I suggest you try doing a more scientific ABX comparison: Take a short segment of one song that you are SURE sounds different. It should be about 30 seconds or less. Because you are repeating the identical segment over and over you avoid issues of switching in the middle and because it is short enough your acoustic memory should be good enough to tell the difference. Some authorities say that 10 seconds is the longest you should ever loop to retain absolute acoustic memory, but your mind will go crazy with a 10 second loop! Which means that the 30 second loop is going to be a little less reliable for your brain to keep track of, but if you do enough trials you should be reliable.

If at the end of all this (and good luck with it... it's a pain in the ass to prove) then you still end up with the various theories as to why they may sound different and no absolute proof of it. I go with the theory that the faster the computer, the less the difference, but it still boggles the mind.

Personally I've never heard a difference between wav and flac, ever, but I've imagined I have hundreds of times  :-).

[jmone]

MC has all the potential of a great ABX platform as you can configure Zones to have all sorts of different DSP processing.  All that is missing is the ABX engine itself to allow user to switch playback between each zone and record the "votes".

[JimH]

... someone did a blind test at conventions, using many different subjects.   What they discovered is that the faster the computer, the less people were able to tell the difference.

What that indicates is that what dean70 said earlier in the thread is correct - it is likely software-induced jitter.

Please cite the article that has the data.

I don't want JRiver's forum to be used to spread misinformation.  I'm probably going to close this thread.

[InflatableMouse]

http://www.audioasylum.com/forums/pcaudio/messages/7/70925.html

Just quoting the article that I could find, doesn't mean I agree with it.

[6233638]

That suggests to me that there was something else going wrong in their setup, such as using the onboard sound card. (they absolutely do introduce noise when the CPU is busy)
And it doesn't sound like this was ABX double-blind testing - more like single-blind.

[Efjay]

If interested, a blog post comparing flac to wav: http://www.computeraudiophile.com/blogs/mitchco/flac-vs-wav-vs-mp3-vs-m4a-experiment-94/

[dtc]

Yes noticed the same thing myself (execpt with aiff format). There is a slight improvement in sound quality. Even though they are both lossless formats, I believe it is related to software induced jitter - the souce bitrate jumps all over place with flac files, but is constant with aiff or wav.

Just what is "software induced jitter" when you are using async USB where the DAC or usb converter is controlling the timing? JRiver fills the buffer and the aysnc USB controls how it comes out and thus the jitter. I might buy (maybe) that on a very overloaded system, the timing of filling the buffer may potentially vary slightly for wav and flac, but with a asysn USB DAC I do not see how it could have any effect, unless the buffer empties, which I really doubt. I am not trying to be argumentative, I just do not understand the mechanism for "software induced jitter" to affect an async usb DAC.

[fitbrit]

I'm sure nobody's mentioned that MC could benefit from an ABX capability yet.
First! 

[jmone]

You are a genius!

[psam]

Here is a simple test for everybody.

I picked up one of my CDs and ripped a track EAC losless.  Came up with a wav file size 73,078 kb.

I converted it this wav to APE and FLAC. The FLAC file was 25,395 kbytes and the APE file came out 23,781 kbytes. Note that the two files are not the same size.

Then, I converted both files back to wav. In both cases the resulting wav file was size 73,082 kbytes.  This is exactly 4kbytes larger than the original wav file, because of the standard 4k bit padding of the previous conversion, no audio data in there at all.

After that, I used two software programs -KDiff and WinMerge available for free use on internet - to compare the three files. Hex and binary comparisons mind you.

The two wav files resulting from the conversion of the APE and FLAC files were identical to the bit.

This file was almost identical to the original wac file. There were only three differences.

1) The 4k bit padding containing meta-data , non audio.

2) One byte in the very beginning of the file (8th byte if I remember correctly), it looked like a header part and

3) One byte in the very ending of the file, again obviously a file termination checksum.

So , this makes me certain for two things:

1) If indeed I can hear differences between the same APE and FLAC encoding of a wav file, there must be something wrong with the player because both files contain absolutely the same data, to the bit.

2) If I can hear differences between the wav and the FLAC files, I must have Spock's ears... They can spot a difference of two bytes  in seventy three million ! 
Hell, I'll send a CV to Stereophile even!!! 

[kstuart]

Psam - you wasted your time - everyone is already assuming that the files are identical to the bit.

Personal Computers are designed to do things while you wait for the results.

If you are computing the best flight path for the Space Shuttle, it takes a certain amount of time.  If your processor is twice as fast, it will take perhaps close to half the time.

Playback of audio and video files occurs in real time.  PCs were not originally designed to do things in real time.

You can have two bit-identical audio files, but if the playback of one occurs slightly differently, then there will be a sound difference.

This is obvious in video playback.  I can take the same mkv video file, and play it back on two different PCs, and on one, there is tearing and on the other PC, there is no tearing (horizontal lines).  It's the exact same file each time, but different playback.

It should not be so surprising that this occurs with audio playback.   And, like the video file, if you read InflatableMouse's link, you will see that the problem goes away with a faster computer.

[6233638]

Playback of audio and video files occurs in real time.

But not decoding. Decoding/playback of most formats should be trivial for most computers these days.

This is obvious in video playback.  I can take the same mkv video file, and play it back on two different PCs, and on one, there is tearing and on the other PC, there is no tearing (horizontal lines).  It's the exact same file each time, but different playback.

This is broken playback of the file though, and would not be a "bit perfect" representation of the source material.

It should not be so surprising that this occurs with audio playback.   And, like the video file, if you read InflatableMouse's link, you will see that the problem goes away with a faster computer.

The link doesn't prove anything - there are important details left out. I will absolutely agree that if you are using the analog output of an on-board soundcard, that CPU usage has an audible effect, and if you are using a player that is "streaming" the data from the disk rather than decoding in advance and playing back from memory, you could potentially hear a difference between a losslessly compressed format and an uncompressed format, or even just an uncompressed format when the system is being stressed vs not.

I don't know whether that would also apply to the digital output of an onboard soundcard. (most likely not) If you are using a good soundcard or external DAC though, I don't believe that there would be any audible difference.

[dtc]

Psam - you wasted your time - everyone is already assuming that the files are identical to the bit.

Personal Computers are designed to do things while you wait for the results.

If you are computing the best flight path for the Space Shuttle, it takes a certain amount of time.  If your processor is twice as fast, it will take perhaps close to half the time.

Playback of audio and video files occurs in real time.  PCs were not originally designed to do things in real time.

You can have two bit-identical audio files, but if the playback of one occurs slightly differently, then there will be a sound difference.

This is obvious in video playback.  I can take the same mkv video file, and play it back on two different PCs, and on one, there is tearing and on the other PC, there is no tearing (horizontal lines).  It's the exact same file each time, but different playback.

It should not be so surprising that this occurs with audio playback.   And, like the video file, if you read InflatableMouse's link, you will see that the problem goes away with a faster computer.

Most people will agree that a PC is not a very good real time system. But with a aysnc USB DAC it does not have to be. With an async USB DAC, the DAC controls the timing of getting the data from the computer and it is dedicated to that task. It is a dedicated real time system.  All the computer has to do is keep the buffer from emptying. Unless you PC is absolutely swamped that is a pretty easy task. With today's systems, there is no reason a PC should be running at more than 5% or 10% of max. when playing music, unless you are using a lot of DSP options.

Some people claim that the electrical noise in the PC is somehow making its way into the converter/DAC and changing its electrical properties which causes problems and that the noise is greater when decoding compressed formats. However, I have not seen any convincing evidence that this is anything but speculation. Plus more and more DACS/converters have galvanic isolation which should take away any noise issues.

If people are actually hearing a difference between wav and flac files on a modern PC with a async usb DAC, then  it seems to me that the converter/DAC is not doing what it is suppose to do and/or the PC is very overloaded.

If someone has a explanation for timing errors/jitter  when using an async USB converter/DAC, I am all ears.

 

[pschelbert]

flac and wav: the only difference is that flac eats around 30-50% less hard-disk space.
For a computer, flac is the way to go.
There is NO sound difference (its equal bit for bit).
wav is useful if you device cannot read flac.

Sound in converters.
Yes if you have bad sound-cards its an issue. I had creative, you hear each access to the hard-disk in the speakers!

Now I have external audio-interfaces, for that reason and you get more flexibility than always these PCI cards with driver issues. Anyway shielding in a computer is difficult (narrow and power-supply noise suppression is difficult).

With my external ones (Motu828mk3 hybrid and RME Fireface UFX) I have no issue whatsoever with computer noise anymore.

Peter

[Neco]

You can have two bit-identical audio files, but if the playback of one occurs slightly differently, then there will be a sound difference.

This is obvious in video playback.  I can take the same mkv video file, and play it back on two different PCs, and on one, there is tearing and on the other PC, there is no tearing (horizontal lines).  It's the exact same file each time, but different playback.

There are probably at least half a dozen reasons why that could happen, and I somehow doubt all of them are absolutely related to the speed of the computer.

[DarkPenguin]

If you're using a DAC that relies on the PC clock then, yeah, all kinds of things can happen to the sound.  Why would you use a DAC that relies on the PC clock?

Psam - you wasted your time - everyone is already assuming that the files are identical to the bit.

Personal Computers are designed to do things while you wait for the results.

If you are computing the best flight path for the Space Shuttle, it takes a certain amount of time.  If your processor is twice as fast, it will take perhaps close to half the time.

Playback of audio and video files occurs in real time.  PCs were not originally designed to do things in real time.

You can have two bit-identical audio files, but if the playback of one occurs slightly differently, then there will be a sound difference.

This is obvious in video playback.  I can take the same mkv video file, and play it back on two different PCs, and on one, there is tearing and on the other PC, there is no tearing (horizontal lines).  It's the exact same file each time, but different playback.

It should not be so surprising that this occurs with audio playback.   And, like the video file, if you read InflatableMouse's link, you will see that the problem goes away with a faster computer.

[compcond]

JRmark: 3573
DAC is Benchmark DAC1 HDR.

Tried listening again.  Same result.  .wav sounds significantly better.

Not blind test but I can't be this crazy

[dean70]

Certainly an interesting topic. Might see if it is anything that can be measured with RMAA tests.

There have been some interesting points brought up here

[bulldogger]

JRmark: 3573
DAC is Benchmark DAC1 HDR.

Tried listening again.  Same result.  .wav sounds significantly better.

Not blind test but I can't be this crazy

I experienced the same thing on a friends computer who was not using JRiver, .wav sounding significantly better. I don't hear a difference with my own set-up however.

[dtc]

You need to do a blind test.  I know once I started doing blind test, I found that I could not reliably tell the difference.

Enlist a friend to run the test. Label 20 pieces with 10 saying wav and 10 saying flac and put them in a hat. Have your friend pick out 10 of the papers one at a time to determine a play order.  Have him write down the order. Pick a track you think you hear the differences on. Have you friend run 15 to 30 seconds of the track. Pause for 5 seconds while you write down whether you think it is wav or flac. Then have your friend start the next selection of the same track. If you have a Android device have your friend use Gizmo, or use library server on a laptop to make the selections, just to be sure you cannot see what your friend is doing. That is not a rigorous abx test, but it should be instructional.

If you want to make it more interesting try a flac file at 0 compression.

What is your cpu usage when playing? I believe you are using the usb to the Benchmark. Correct? It has excellent jitter rejection, so jitter should not really be an issue.

[sla]

There is no difference between WAV and FLAC if both formats are handled properly. and JRiver handles them properly.

Look how about this: can WAV files sound differently in the morning then in the evening?

the answer: the system/speakers will output the same sound but the way we perceive the sound may be different in the morning then in the evening [being tired/ versus relaxed, etc...]

that is why there is a placebo testing for medicines. Just if you assume that WAV sounds better it will for you...

[bulldogger]

that is why there is a placebo testing for medicines. Just if you assume that WAV sounds better it will for you...

  My expectation was that there would be no difference. I resisted the comparison. I was convinced my friend was wasting my time. Had he not insisted, I would have never listened.  So if an expection effect like placebo was working, the results should have been I heard no difference. Blind test are a good idea. Many assume that placebo save you from hearing differences that are not really there and it can.  However, placebo is about expectation and that can go in any direction. If you expect to hear not difference, that expectation can also work so that you do not hear any difference.

[DoubtingThomas]

Sorry guys... if the playback DAC engine gets the same bits for wav and flac and ape (and they do, it's been proven) then there is no difference between formats.  There are so many real things one can do to improve sound rather than focusing on wav/flac/ape...  If you hear a difference, then you want to hear a difference.

I used to sell high end audio back in the late 70's...  I could make any system sound better than another with even the slightest difference in levels..  we sold what made us the most money...

[kstuart]

Babyface wrote: "Sorry guys... if the playback DAC engine gets the same bits for wav and flac and ape (and they do, it's been proven)"

You are missing the point (just like psam missed the point).

It is not which bits, it is when.

And again, InflatableMouse's link has results of a blind test.

But when you say " I'll believe it when you do a blind test. " you really don't mean it.  You prefer a world where everything can be measured, and people in authority are competent and know what they are doing.  Dream on.

PS  You mentioned "the playback DAC"... the post with the results of a blind test is posted by a DAC designer.

[6233638]

You are missing the point (just like psam missed the point).

It is not which bits, it is when.

The Benchmark DACs reclock all incoming audio so that it eliminates jitter - this is not a factor.

And even on devices that are susceptible to jitter, it is typically well below the audible range, especially when music is playing.

And again, InflatableMouse's link has results of a blind test.

A single-blind test which does not go into detail about the playback system. The way it is described, it sounds like they are using the analog line-out of the MacBooks - a known bad source. They say the systems are dual booting Vista/OS X and running JRiver/iTunes - but not how the tests were performed.

They rely on the "If you have an average system then none of this applies." line - a pretty clear indicator that this has been an "audiophile" conducted test, and not a scientifically controlled test.
The audiophile argument for anything is "well your system isn't good enough" or "your hearing isn't good enough" etc.

[jmone]

...and we are off to the races, in same two different buses that these threads always generate....

I'll believe an ABX engine would sort this out.  I also think that when requests for another audiophile feature is asked for that JR put up an ABX web portal or even better an on line MC library to test such claims before they are added.

[Vincent Kars]

By design PCM audio is about samples + sample rate.
If you alter the sample (any type of DSP) you get another sound.
If the sample rate fluctuates a little (jitter, the digital audio equivalent of wow and flutter) it might become audible.

This is where the concept of software induced jitter is about.
As all what the software does translate into activities performed by the hard ware and as a consequence might generate EMI, RFI, ripples on the power, some noise on the ground plane, etc  and this might creep into the DAC affecting the DA conversion.

An obvious first: this has nothing to do with the bits.
This can be easily proved: play a WAV, play a FLAC (with identical content of course) and record the bits e.g. using the SPDIF out.
This yields zero difference.
A nice example are the experiments by Mitcho: http://www.computeraudiophile.com/blogs/mitchco/flac-vs-wav-part-2-final-results-155/

I have run a RMAA one day on one of my laptops. One run with the PSU in place and one with battery power only. One can clearly measure the differences:
http://thewelltemperedcomputer.com/Intro/SQ/SoftwareInducedJitter.htm

Recently Archimago did some very interesting experiments.
He run the J-test (jitter test) using various protocols (adaptive USB, async USB, Toslink) but also changed the system load form “normal to 100% CPU/GPU
http://archimago.blogspot.ca/2013/03/measurements-adaptive-aune-x1.html
http://archimago.blogspot.ca/2013/03/measurements-hunt-for-load-induced.html
He could find only some very marginal differences.

My thoughts:
As PCM is sample + time step, you have to take both into account, focusing on the bits only leave half of what PCM is out of the equation.
It is possible that the electrical activities going on inside a PC affect sound quality.
It is probably highly system dependent.
It is certainly not a one to one relationship.
A measurable difference is not necessarily also an audible difference.
Reported differences in sound quality are often perceived differences.
Perceived differences even those “night and day”, “blown away”, “not by a small margin” has the nasty habit to disappear completely when tested unsighted.

My motto
Trust your ears (what else can we do?)
Mistrust your perception (what else can we do?)

[rayooo]

As with all of these types of objective vs subjective threads, all 12 or so million of them since the beginning of "high-end audio",
science has little (actually Zero in my opinion) to do with it.  99.9% of people are in one camp or the other, and since the first thread on the first day, Post #1. A grand total of .7 persons have changed their minds based on threads like this one.

This thread could easily be about Silver VS copper VS Gold VS Rare-earth quantum adjusted wire, or high end fuses, or damping rocks.  Just change a few words here and there and you've got it.

These are fun to watch now and then, this one WAS also.

OBTW, can anyone recommend a 1' long good sounding USB cable?  My budget is only $800 for this, I'm hoping to add a bit of warmth to my system.
And can I use the same USB cable with FLAC and WAV?  I hoping to not have to swap USB cables based on what audio format I'm playing at a given moment.  

[bulldogger]

The opposite of placebo effect is the nocebo effect. http://www.time.com/time/magazine/article/0,9171,1931727,00.html   So if you don't determine a difference, it can also be because you have decided you do not want to hear one if the test is not double blind. There is a reason for the double blind protocol. When the experimenters. themselves, know which sample is which, it also effects the results.