HDtracks now offers AIFF, ALAC, WAV, & FLAC

[jackman]

One more note, I don't intend to plug one manufacturer, but the new NAD M50 server, when ripping a cd, rips it to ram memory and then converts to flac at its leisure. They are adding a wav ripping option in response to dealer and early customer requests, it will be available very soon in a software update. Also, and I'm not sure I'm remembering a conversation accurately, but I think one of the guys working on the software development told me that they found ripping to ram better sounding than real-time conversion.

[Julf]

Julf, can you point me to a reference implementation of the flac codec.

 

FLAC on sourceforge

 

A big issue I'm finding with computer based audio is everything seems to make a difference.

 

And what I find great is that in a well-designed system there are so many things that *don't* make a difference.

 

In an analog system, pretty much everything affects everything else, and errors/distortion add up.

 

In a (again - well designed) computer system, the input of the DAC is pretty much the first place where errors/distortion starts to show up.

 

If something is corrupting the flac decode process, we owe it to ourselves to try to figure out how to prevent that.

 

Agree 100%!

[Julf]

One more note, I don't intend to plug one manufacturer, but the new NAD M50 server, when ripping a cd, rips it to ram memory and then converts to flac at its leisure.

 

Pretty much all the rippers/encoders running on Linux I am familiar with use virtual memory as their buffer, and as long as you have enough RAM, it will all stay in RAM.

[drwho]

jackman

 

Your trouble with flac must be an implementation issue. I invite you to do the following test. In Ubuntu it is very easy to get the reference flac encoder. You can make an Ubuntu live USB stick that you can boot into without even installing Ubuntu. Once booted, open a terminal and install flac by typing

 

sudo apt-get install flac

 

Insert another USB stick into the computer with your wav file. In my example I used "192k_24bit.wav". You will need to change directory into the inserted USB stick. Type

 

cd /media/[press tab key to list your inserted USB stick and type the first letter of the stick, then tab again, then enter]

 

My usb stick was

cd /media/68CA-62A1

 

first convert the wav file to flac by typing

flac -0 192k_24bit.wav

 

//a file called 192k_24bit.flac will be created. Next convert the flac file back to wav

flac -c -d 192k_24bit.flac > output.wav

 

//Finally compare the original and converted files with the md5 hashing algorithm by typing

md5sum output.wav 192k_24bit.wav

//its output is....

eb6135cd9fde736a779a864a95fac41e output.wav

eb6135cd9fde736a779a864a95fac41e 192k_24bit.wav

 

You will notice that the md5sums are identical. Not one bit is different!

[Cycleman]

I am still occasionally buying downloads from sites like HDTracks, but only after having verified the material is bona-fide hi-res.

 

And how do you perform this pre-purchase verification?

 

Dave, who is new to high-res music files and just bought a so-called hi-res album from HDtracks today and feels like he got burned

[Julf]

And how do you perform this pre-purchase verification?

 

Badly, as can be judged by this thread where I got burned again. Usually I wait for someone to post a spectrum plot, but for 44.1 kHz material there wouldn't be any high frequency stuff anyway - only hi-res part was the 24 bit depth. Except one third of the package I got was zeroes...

[vinylphile]

WAV vs. FLAC - this topic is always an interesting battle. My personal approach is to store everything as AIFF on my MAC. The only disadvantage I see is the greater storage space required vs. ALAC. The potential advantage is greater fidelity of an uncompressed format. I fully realize that this is an "unscientific" approach but I have NOT been convinced that we have absolute confirmation that lossless and uncompressed sound identical under all circumstances. Storage is cheap - I can store a thousand albums or more in AIFF on $200 worth of hard drive space. I don't want to waste precious listening time with tedious A/B testing. Why sweat it?

[Julf]

I have NOT been convinced that we have absolute confirmation that lossless and uncompressed sound identical under all circumstances.

 

While we have what is as close as modern science can get to absolute confirmation that storing the music using a lossless compression doesn't change the contents in any way, we indeed do not have anywhere near absolute confirmation that actually unpacking the file format in real time while playing it would not cause some sort of audible effects in the player/DAC combination - but we also don't have confirmation that the additional I/O (disk or network) caused by uncompressed formats doesn't have effects. Thus uncompressed music could sound better or worse than losslessly compressed music in a specific setup - and the only way to know for sure would be proper, controlled blind ABX listening.

[vinylphile]

While we have what is as close as modern science can get to absolute confirmation that storing the music using a lossless compression doesn't change the contents in any way, we indeed do not have anywhere near absolute confirmation that actually unpacking the file format in real time while playing it would not cause some sort of audible effects in the player/DAC combination - but we also don't have confirmation that the additional I/O (disk or network) caused by uncompressed formats doesn't have effects. Thus uncompressed music could sound better or worse than losslessly compressed music in a specific setup - and the only way to know for sure would be proper, controlled blind ABX listening.

 

Agree that if there is any difference it would likely be due to file conversion.

 

However, disagree on most of the rest. I can't envision any logical scenario where keeping all files in uncompressed formats could potentially lead to worse sound quality. It is difficult for me to think of a scenario where converting the file twice would lead to any improvement. My experience in audio and music has led me to believe that the less manipulation, the better. In my opinion, the best that FLAC or ALAC could possibly sound is AS GOOD AS the native uncompressed format. Luckily since storage is so cheap, it doesn't really matter, at least in my opinion.

 

In terms of double-blind testing, this is another area where I disagree. I do think it can be a valuable tool however it isn't the be-all-and-end-all. Yes, double-blind tests can remove certain biases and placebo effects - however they have problems of their own.

 

If product or file A sounds the same as product or file B in a double-blind ABX test, the only possible conclusion is just that - that product or file A sounds the same as product or file B in the test. To then say they sound the same under all conditions is extrapolation and a bit of a leap of faith, IMO.

 

Again, I have participated in AB audio testing and do not find those listening conditions can necessarily be extrapolated to real world listening. I'm not saying that such testing is without merit - only that its rabid supporters never seem willing to accept the testing methodology's limits.

[Julf]

However, disagree on most of the rest.

 

Sure - I totally get your view. But...

 

I can't envision any logical scenario where keeping all files in uncompressed formats could potentially lead to worse sound quality. It is difficult for me to think of a scenario where converting the file twice would lead to any improvement. My experience in audio and music has led me to believe that the less manipulation, the better.

 

OK, let's start with the latter part - I agree with you that less manipulation is better if we are talking about manipulation (such as resampling and processing) that actually changes the content. But if we are talking about encoding the music in a lossless form, and then decoding it back, there is *no* change. You can easily verify that (and it is an useful exercise) by running a script (overnight) that takes a raw wav file, endodes it into flac, decodes it back to wav, re-encodes it into flac, and so on for at least 10,000 times. Then listen to the result, and compare it to the original one. I did that in one of my listening tests - no difference detected.

 

As to situations where uncompressed might lead to worse sound quality - if there are audible differences, they are not coming from any changes to the bits themselves, as we have just discussed, so the differences would be caused by jitter and noise affected by the CPU load caused by the decoding operation. Unfortunately shifting the bits off a hard disk and over a network causes CPU load too, especially in the embedded (usually ARM-based) embedded CPUs used in network players/streamers, as the network is handled in software rather than by a separate network interface processor. This means that the additional I/O load from uncompressed data might actually cause more CPU load than the decoding of the compressed format. I have seen actual measurements from a Logitech Squeezebox showing higher jitter with uncompressed data compared to flac. Whether that higher jitter is audible is another issue altogether.

 

In my opinion, the best that FLAC or ALAC could possibly sound is AS GOOD AS the native uncompressed format.

 

As you see, that is not always true.

 

In terms of double-blind testing, this is another area where I disagree. I do think it can be a valuable tool however it isn't the be-all-and-end-all. Yes, double-blind tests can remove certain biases and placebo effects - however they have problems of their own.

 

Sure, I think we all agree there are problems. It is just that, as I keep saying, it is a bit like democracy - it is a flawed system, but is the best one we know. I haven't seen critics of double-blind tests propose any equally reliable and verifiable methods that would deal with the perceptual biases. We all know that any improvement we make to our systems *will* sound better, and the solid, shiny and heavy $15,000 box *will* sound better than the $200 plastic box...

[vinylphile]

Sure - I totally get your view. But...

As to situations where uncompressed might lead to worse sound quality - if there are audible differences, they are not coming from any changes to the bits themselves, as we have just discussed, so the differences would be caused by jitter and noise affected by the CPU load caused by the decoding operation. Unfortunately shifting the bits off a hard disk and over a network causes CPU load too, especially in the embedded (usually ARM-based) embedded CPUs used in network players/streamers, as the network is handled in software rather than by a separate network interface processor. This means that the additional I/O load from uncompressed data might actually cause more CPU load than the decoding of the compressed format. I have seen actual measurements from a Logitech Squeezebox showing higher jitter with uncompressed data compared to flac. Whether that higher jitter is audible is another issue altogether.

 

Guess it would depend on your particular situation. In mine there is no network involved in transferring the data - other than when I actually download the track, which I actually usually do in FLAC format before converting it to AIFF using Max. Of course one of the reasons I do this is because of issues with FLAC on a Mac - but I choose AIFF instead of ALAC. Once that step is achieved, no network is involved - I store the files on an external 2TB Thunderbolt drive (VERY fast!) and play everything back through cache memory - so I'm not transferring files in real time on a network but ahead of play via Thunderbolt and any other file manipulation is necessary. So again, to me, it just seems logical that while I would guess ALAC and AIFF probably sound similar to identical, in my situation if there is any benefit to either of them, AIFF would appear to have the advantage. If transferring files is the likely bottleneck then I could see your point.

[vinylphile]

Sure, I think we all agree there are problems. It is just that, as I keep saying, it is a bit like democracy - it is a flawed system, but is the best one we know. I haven't seen critics of double-blind tests propose any equally reliable and verifiable methods that would deal with the perceptual biases. We all know that any improvement we make to our systems *will* sound better, and the solid, shiny and heavy $15,000 box *will* sound better than the $200 plastic box...

 

Again, that's a matter of opinion. Every testing methodology has its pros and cons and yes the randomized, double-blinded, placebo-controlled (if applicable) trial is considered the gold standard in many scientific endeavours. However, a blanket statement that it is the only valid testing methodology in audio is your opinion, not fact. In my experience I find it to be less useful than an extended, unrushed audition in a natural listening environment. Such auditions have produced results both expected and unexpected (i.e. the opposite of my own expectation bias) - whereas I sometimes have difficulty "listening to the music" rather than "listening to the differences in sound" when I'm A/B testing. I personally employ both techniques when evaluating new equipment but find an extended audition more enlightening.

 

In fact, I tend to find A/B testing more tedious than enlightening, which is why I haven't bothered A/B-ing ALAC vs. AIFF in my situation.

[Julf]

Guess it would depend on your particular situation.

 

Definitely.

 

In mine there is no network involved in transferring the data

 

Right - so network traffic is not an issue for you. Disk I/O might still be, and the I/O load is not so much dependent on the speed of the disk drive, but the amount of data to be transferred. I agree that in your case it should not be much of an issue. My point is still that the only way to *know* is by a listening test - you can't categorically say that uncompressed would sound at least as good as compressed. It might, or it might not.

[Julf]

Every testing methodology has its pros and cons and yes the randomized, double-blinded, placebo-controlled (if applicable) trial is considered the gold standard in many scientific endeavours.

 

For good reasons.

 

However, a blanket statement that it is the only valid testing methodology in audio is your opinion, not fact.

 

That is not (quite) what I wrote. What I stated was "Thus uncompressed music could sound better or worse than losslessly compressed music in a specific setup - and the only way to know for sure would be proper, controlled blind ABX listening."

I stand by my statement that without a proper, controlled blind test that removes the possibility of (unconscious) bias and placebo effect, we can not *know for sure*, in the sense of having objective, verifiable and replicable data.

 

In fact, I tend to find A/B testing more tedious than enlightening, which is why I haven't bothered A/B-ing ALAC vs. AIFF in my situation.

 

I agree it is tedious - and hard work. That is why most people don't like to do ABXing. Nobody really enjoys it. But how else can you know that what you think you hear is a result of real, verifiable differences? And just to be clear, that is not a way to argue my case, but a honest question. I would love to be able to avoid the tedium of doing ABXs, but have done enough tests to realize that even if I *know* I hear some difference, once I remove the possibility of unconscious bias the difference often goes away.

 

I am also painfully aware this discussion is turning into Yet Another ABX versus sighted listening thread that has nothing to do with HDTracks...

[vinylphile]

So I prefer to listen in my normal listening environment and recognize potential bias and placebo. You prefer ABX testing and recognize the potential lack of external validity of such testing. We're good!

[Julf]

So I prefer to listen in my normal listening environment and recognize potential bias and placebo. You prefer ABX testing and recognize the potential lack of external validity of such testing. We're good!

 

We are!