FLAC versus WAV

[Vincent Kars]

I don't think that most audiophiles are in the market for humbling experiences.

In 2007 I bought a music server.
I simply ripped a couple of tracks to WAV, FLAC and 256, 192, 128 MP3.
I didn’t hear a difference between WAV and FLAC as expected
With decreasing bit rate the sound was getting worse. Even 256 was obvious down, an apparent loss of inner detail as expected.

I decided to rip to FLAC (lucky me).
On the Hifidelio forum, someone said that there is no audible difference between FLAC and 256. I protested, I had heard these differences myself.
Instead of a flare, we decided on a blind testing.
I supplied a couple of tracks and another guy converted them to various MP3 flavors and converted them back to WAV.
Not a single visible clue (dam…).
Comparing 128 with the original was easy, 192 difficult and with 256 I was completely at loss. The advantage of a unsighted test is that nobody can see that you are cheating.
I tried auCDtect. It turned out to be like me, a failure when distinguishing between the original and a 256.
In the end I was complimented with my almost perfect score (a very lucky guesser obvious) and had to answer questions about how to discern between the original and MP3…….

Lessons learned:
- do your testing sighted, it really improves your score…..
- high bit rate MP3 is like magic, throw out 90% to maintain 99% quality

 

[Phelonious Ponk]

In 2007 I bought a music server.
I simply ripped a couple of tracks to WAV, FLAC and 256, 192, 128 MP3.
I didn’t hear a difference between WAV and FLAC as expected
With decreasing bit rate the sound was getting worse. Even 256 was obvious down, an apparent loss of inner detail as expected.

I decided to rip to FLAC (lucky me).
On the Hifidelio forum, someone said that there is no audible difference between FLAC and 256. I protested, I had heard these differences myself.
Instead of a flare, we decided on a blind testing.
I supplied a couple of tracks and another guy converted them to various MP3 flavors and converted them back to WAV.
Not a single visible clue (dam…).
Comparing 128 with the original was easy, 192 difficult and with 256 I was completely at loss. The advantage of a unsighted test is that nobody can see that you are cheating.
I tried auCDtect. It turned out to be like me, a failure when distinguishing between the original and a 256.
In the end I was complimented with my almost perfect score (a very lucky guesser obvious) and had to answer questions about how to discern between the original and MP3…….

Lessons learned:
- do your testing sighted, it really improves your score…..
- high bit rate MP3 is like magic, throw out 90% to maintain 99% quality

I'm pretty sure I couldn't consistently differentiate 256kbps from lossless either, but I still rip all CDs to lossless. HD space is cheap, why archive with less. I've completely lost my hesitation to buy from iTunes, though, since they went to 256.

Tim

 

[FrantzM]

Hi

I am fairly consistent on my 256 score .. I get them right on foobar ABX' close to 90% of the time with my headphones ... 320 is harder but I do get it once in a while a little above 70% when I know the original CD very well..

I don't care that much anymore with the price of storage plunging down ... I seldom buy mp3 .. Rather I buy the CD or download Hi-Rez High Rez when available ...

 

[garylkoh]

Frantz,

have you done ABX on hi-rez vs 16/44.1? I score below 70% if the quality of the original file is good. The benchmark I use is Russian Roulette - you have a 83% chance of survival. I wouldn't play Russian Roulette, and if the probability of being to tell the difference between high rez and CD is below 83%, I wouldn't play....... although the audiophile in me still compels me to buy the high-rez.

I think more important than the sample rate is 24 bit.

Just in case someone noticed..... I used to swear that I can hear the difference between 44.1 and 96kHz, until another member on this forum pointed me to the SoX resampler.

 

[Bruce B]

I think more important than the sample rate is 24 bit.

I totally agree with this. You can have a 24/44.1 piece and a 16/384 piece and the 24bit will sound better every time. You can pick out the 24bit music 100% of the time after just a few notes.

 

[Old Listener]

I think more important than the sample rate is 24 bit.

Did you mean

"I think more important that the sample size is 24 bit. "

 

[garylkoh]

Did you mean

"I think more important that the sample size is 24 bit. "

Sorry, I was in a hurry. I meant - I think more important than the sample rate being 44.1kHz or 96kHz or 192khz is that the sample size should be 24 bit and not 16 bit."

Having heard Bruce's DSD recorder, I'm also reconsidering that 1-bit at a high enough sample rate is also excellent.

 

[Bjorn]

Just want to let you know that I'm in a process of blindtesting Wav vs. flac files on myself. I'm using J.River Media Center. I'll do the blindtest over a period of time and do several ones. So far I've have contucted two tests. The result is below.
1. 8/10 correct
2. 9/14 correct

Could be pure luck so far. We'll see.

 

[amirm]

Interesting. Let us know how you end up.

 

[mojave]

Isn't WAV also a compression algorithm of some sort? Many view WAV as uncompressed lossless, but in reality it is just a less efficient method of compression.

I found this quote by Matt, the developer of the APE lossless format, interesting:

This is a side topic, but lossless compression is a lot more secure (from an audio fidelity standpoint) than WAV.

APE has multiple redundant error checking systems built into the decoder to ensure the output data exactly matches the input data. I think FLAC probably has something like this as well.

A WAV file can be corrupted and you would have no way to know. This is not just theoretical. Twice I've had bad system memory that corrupted my music (I had backups). With WAV, I would have never known.

 

[amirm]

If we are talking about traditional audio files kept in .wav files, no they are not compressed. If your CD has 44,100 samples/sec, you have just as many in the .wav file.

The error resilience is of little use in our applications. Computer systems don't lose data often (or else your system wouldn't work right). Error correction matters if you are going over lossy networks. But a file going from your PC to your DAC doesn't travel that way. Yes, it is possible to lose data over USB but since the APE file is not sent over USB as such (it is sent as uncompressed samples) even if some data was lost there, the fact that you used APE doesn't help.

Perhaps there is some use in downloading files and knowing if they are corrupt or not but not sure how often that happens or matters.

 

[Vincent Kars]

WAV is part of the RIFF family.
It is a container format so you can put anything inside.
In practice it is set to PCM, uncompressed lossless audio.
What Matt mentioned, corrupted file due to failing system memory, is indeed a cause of data rot. This can happen if you transfer a file.
If APE works like FLAC, on playback the decoder will trow a warning when the content and the checksum differs.
The specs: http://thewelltemperedcomputer.com/Lib/WavSpec.txt

 

[_Wim_]

I find it strange that people can hear the difference between 16 bit and 24 bit. This only makes sense to me is it is a very soft recording (lots bits already wasted). 16 bits gives theoretically a 96 db (24 bits 144db) signal to noise ratio, taken into account you have already minimum 40 db of background noise, you would have to play really loud to hear the most silent bits...

I am aware that background noise is uncorrelated noise, and you can hear correlated signals below this noise floor, but not 40 db down below the noise floor...

From a purely technical standpoint I could imagine it is possible for young people (still a very good high frequency hearing) to hear the difference between 44.1 and 96kHz sampling rates (on a recording with lots of HF details), this because the lowpassfilter on the 44 kHz signal is pretty close the the audible frequency range.

I have tried to hear the difference between 16 bits and 24 bits with a headphone, and I did not succeed (nor did I succeeed in hearing the difference between 44.1 and 96 kHz, so it is maybe my hearing...)

 

[garylkoh]

Wim - it may boil down to the original recording.

Our resident expert has a test here:
http://www.whatsbestforum.com/showthread.php?2938-The-Art-of-Listening-Hi-rez-music

many of us can hear the difference.

 

[RBFC]

Isn't the fact that most systems only reproduce 12-13 bits of resolution one good reason that extra "headroom" is important?

Lee

 

[_Wim_]

I do not see how more bits would create extra headroom per se. The reason that some equipment only has 12 to 13 bits of resolution, is not the DAC chip. So placing a DAC chip with 24 bits in that same position will not solve anything. The only very good reason I see for 24 bits, is in the digital domain itself (recording environment), where there are lots of complex signal calculations (filters, special effects, echo, expansion, limiting...) in series before the final result is available. If these where perfomed with 16 bits only, small calculation errors would add up in the end result and would become audible. With 24 bits these errors are much much smaller so this does not become a problem.

An (silly) example: say I have 2 filters in series and a 16 bit recording. This first filter creates a 12db (2 bits) dip from 1 to 4 kHz. The second filter creates a 12db (2bits) boost from 2 to 4 kHz, essentially undoing the first filter between 2 and 4 khz. If I did not have any headroom (low level bits with zeros in it), the 2 lowest bits of resolution between 2 and 4 kHz would be lost when these 2 filters where applied in series. The 24 bits format fixes this problem.

 

[RBFC]

Agreed for sure that the recording side of things is of primary importance.

Lee

 

[DonH50]

To me, 24 bits makes a big difference in the studio to provide extra headroom for recording and mastering. I remain skeptical of its need in the final product but there are a lot of variables, not the least of which is DAC linearity (which may be much better in a 24-bit device even at the same level as a 16-bit device) and of course there's always the resolution noise floor argument. given that and other real-world effects, a 24-bit DAC might sound better even though on paper there should be no practical difference going from 16 to 24 bits.

All imo, fwiwfm, my 0.000001 cents, blah blah bah - Don

 

[Orb]

To me, 24 bits makes a big difference in the studio to provide extra headroom for recording and mastering. I remain skeptical of its need in the final product but there are a lot of variables, not the least of which is DAC linearity (which may be much better in a 24-bit device even at the same level as a 16-bit device) and of course there's always the resolution noise floor argument. given that and other real-world effects, a 24-bit DAC might sound better even though on paper there should be no practical difference going from 16 to 24 bits.

All imo, fwiwfm, my 0.000001 cents, blah blah bah - Don

Don, whats your experience relating to recordings sold as 24bit?
I am wondering how many actually use real 24 bit information of the recording and is not just random noise,etc, or a transfer of of 16bit recording.
Are all studio masters actually done in 24 bits these days, and if so just curious any idea when studios started to do this.

BTW not sure if you have been following the following thread, ignore the persistent same few doing the usual arguing against JA and there is a lot of food for thought about 24bit recordings from others including JA in the thread I think from the page onwards linked.
http://www.hydrogenaudio.org/forums/index.php?showtopic=86649&st=275
Cheers
Orb

 

[garylkoh]

fwiwfm, imho, the 24-bit 44.1kHz Beatles USB Apple sounds absolutely wonderful to me!!