High Res Files Comparison

[Ammonite Audio]

The 2L High Res Download Testbench at http://www.2l.no/hires/index.html provides an opportunity to listen to identical musical performances, using different file formats, eg 96kHz FLAC, 96kHz WAV. I have long harboured a suspicion that streaming uncompressed WAV offers better musical reproduction than lossless FLAC, mainly because of the lightened processing load on the host computer (I do accept that the bits, if not the timing of those bits, are the same once expanded).

Using Foobar via my Paul Hynes tweaked HiFace Evo into the Lyngdorf amp, there is a very clear difference between 96kHz FLAC and 96kHz WAV (the amp's display clearly showing the correct sampling rate in both cases). The 2L performances are really rather good, and on the "Mozart: Violin concerto in D major - Allegro
Marianne Thorsen / TrondheimSolistene" track, the sense of natural space, tonal colour and air that the WAV file provides is not at all matched by the FLAC file which sounds rather constrained in comparison.

Fortified by this finding, I tried to play the 96kHz WAV files through J River Media Center 16, but for some reason it does not seem to accept WAV files other than 16-bit. It does play the 96kHz FLAC files, though, and here the standard of reproduction is more or less on a par with the WAV files via Foobar. I hope that I can find a way to play the WAV files through J River, so if anyone has any wisdom to share, I'd be happy to listen (no joy with the J River Wiki or Forums).

So, the interim conclusion here is that, depending on the PC hardware and media software used, there is a clearly heard difference between FLAC and WAV. Specifically, J River Media Center is significantly better than Foobar and well worth the $50 licence cost. FLAC does offer better tagging etc and is convenient, so there are of course other factors to take into account.

[StanleyB]

I have mentioned the same thing on a few occasions, but my findings have constantly been dismissed. But encouraged by your findings, I'll consider my search for a revealing amp and pair of speakers now resolved :).

[Ammonite Audio]

Martin Colloms has stated pretty much the same thing on the HiFi Critic Forum, and then had to deal with the "bits are bits" objectivists! I am happy to trust my ears, but I do wonder how good high-res WAV could be through J River Media Center, if I can find a way to make it work.

[Vincent Kars]

Bit dark as the 24 bit FLAC does play.
In JRiver in DSP studio, if you set the output format to "source bitdepth" does that help?

[Ammonite Audio]

Bit dark as the 24 bit FLAC does play.
In JRiver in DSP studio, if you set the output format to "source bitdepth" does that help?

I'm not in front of the PC at the moment, but I'm sure that's how I have it set up. In the supported File Types, J River MC16 lists only 16-bit WAV, which I take to be the problem.

I do need to finish my bespoke Media PC build, which will have to use J River MC17, so maybe that will cope with high res WAV. We shall see (or rather, hear)!

[Darren]

How fascinating. So FLAC ain't lossless after all?

[alfie2902]

How fascinating. So FLAC ain't lossless after all?

Darren, I don't think Hugo or anyone is saying FLAC files are not lossless, they can be measured to prove they are lossless infact.

What's being brought to question is, does the unpacking of FLAC files have a detrimental effect on sound quality, caused perhaps by the extra processing power needed whilst unpacking them or added timing issues. There seems to be a growing number of people who claim to hear improvement to sound quality using WAV files.

Interestingly the media player 'JPlay' reduces processor load & to my ears sounds better than any file player I've so far tried, it's a bit basic in operation but to my ears offers an improvement over any others I've used. It might be worth a try with your WAV's Hugo!

[bobbasrah]

Darren, I don't think Hugo or anyone is saying FLAC files are not lossless, they can be measured to prove they are lossless infact.

What's being brought to question is, does the unpacking of FLAC files have a detrimental effect on sound quality, caused perhaps by the extra processing power needed whilst unpacking them or added timing issues. There seems to be a growing number of people who claim to hear improvement to sound quality using WAV files.

Interestingly the media player 'JPlay' reduces processor load & to my ears sounds better than any file player I've so far tried, it's a bit basic in operation but to my ears offers an improvement over any others I've used. It might be worth a try with your WAV's Hugo!

I hear no difference between WAV and FLAC, and suspect it is down to the power of the respective computer and player software. I DO admit I have heard the difference quoted on less powerful devices and platforms previously, but that is a matter of implementation, not the packaging itself. Heresy in some quarters it would seem.....

[Tony Moore]

+1.

I can't tell the difference between WAV and FLAC on Squeezebox receiver (via I2S) with my current setup.

However, I _can_ tell the difference between versions of the firmware in SB receiver and SB3. I use a very old version of firmware as all the newer ones sound inferior to my ears. I have no idea what they did between firmware version but it seemed to me that it might have affected the FLAC decoding perhaps. I've not done any specific A/B comparisons though so can't go beyond wild speculation...:eyebrows:

[StanleyB]

I hear no difference between WAV and FLAC, and suspect it is down to the power of the respective computer and player software.
How powerful should the computer be? 10GHz?
As for the software side of things: all these audio files have a strict encoding/decoding .dll or equivalent format that would have been written, and even patented, by the original inventors. The operation interface software has to either use the codec from the file format inventors, or stick to it very closely. Otherwise there could be trouble.
I happen to know this part of the problem with codecs when I was busy designing my own mp3 player (MPress3) in the 90's. The Frauenhouser institute was quite adamant that I used their own codec.

[bobbasrah]

How powerful should the computer be? 10GHz?
As for the software side of things: all these audio files have a strict encoding/decoding .dll or equivalent format that would have been written, and even patented, by the original inventors. The operation interface software has to either use the codec from the file format inventors, or stick to it very closely. Otherwise there could be trouble.
I happen to know this part of the problem with codecs when I was busy designing my own mp3 player (MPress3) in the 90's. The Frauenhouser institute was quite adamant that I used their own codec.

As I said Stan, "I suspect", but do not know why the once clear difference is no more, but the hardware and OS were the obvious deviations between the respective devices. Age is also perhaps the other obvious difference, as I cannot tell between 24/96 and 24/192 over usb to the DAC. 10GHz?
As I said "Heresy in some quarters it would seem..... "

[Werner Berghofer]

Robert,


as I cannot tell between 24/96 and 24/192

are you sure you can tell the difference between 16/44.1 and higher rates? If you (and anybody else interested in this topic) have ten minutes to spare I highly recommend reading → this thread (http://www.head-fi.org/t/415361/24bit-vs-16bit-the-myth-exploded) on Head-Fi.

Werner.

[ZebuTheOxen]

As Stan rightly points out, the processing and output software are far more significant in influencing the sound quality coming out of a computer.

Even the most mediocre of processors on the market (Think very low powered Intel Atom processors, 10 / 15 year old desktop systems) will handle 192/24 processing quite easily.

Comparing the output of a WAV file with a FLAC file will reveal no differences before it meets the preamp / output stage of the system. Beyond that, there are many, many layers of obfuscated code which could be interfering.
This is why WASAPI / ASIO / Kernel Streaming are so strongly recommended.

I have the Windows 8 developers kit at home, I'm trying to work out if it's worth making a stripped down image of it, purely for hi-fi PC purposes. At the least I am to make a guide to the process, but I'm quite strapped for time at the moment, who knows! :)

[bobbasrah]

Agreed Tom, but is the clear audible difference being reported between a WAV and FLAC file not possibly something to do with the environment in which the FLAC is decoded, if the outcome is different elsewhere for essentially the same process?
With different platforms, configurations and software implementations, there are a lot of variables, but the file format is being held here as the sole cause, which would only make sense were it to manifest universally.

Werner - I was referring only to the 24 bit files I have that were created from a vinyl transcripton to test 24/96 and 24/192 formats. The difference between the LP and digital formats was minimal but present, the difference between digital frequencies was non existent. I did not have the CD version, but I regret not doing a 16 bit sample as it would have been an interesting comparison at the time.
That being said, I have had a few CDs previously which were poor quality compared to the LP, so comparing the CD to the 24bit from the LP would not have necessarily been honest.

[ZebuTheOxen]

Agreed Tom, but is the clear audible difference being reported between a WAV and FLAC file not possibly something to do with the environment in which the FLAC is decoded, if the outcome is different elsewhere for essentially the same process?
With different platforms, configurations and software implementations, there are a lot of variables, but the file format is being held here as the sole cause, which would only make sense were it to manifest universally.
It is possible for the FLAC environment to vary. The licensing of the libFLAC API (decoder) is basically lip service to the creator and foundation allowing it - http://flac.sourceforge.net/license.html

As a result, there is flexibility for the device engineers, but also the opportunity of non-adherence to the reference implementation. The only way to guarentee this would be to ensure the device specifically implements libFLAC unedited and complete.

Beyond that, it is possible for all layers of the computer's software to influence the sound. Without a controlled bit stream coming straight from the system, who knows what could be happening.

It is slightly jarring for end users, but importantly it does not effect the FLAC container itself (and it's use as an archival format), merely it's playback.