FLAC vs. WAV (a listener's perspective)
Posted by: gary1 (US) on 07 March 2010
OK, here we go.
I downloaded a 24/96 album from HDTracks by Peter Berstein called Monk. Really good album BTW, his take from a guitar trio perspective on the music of TS Monk.
After downloading the album, I then converted it to WAV using Dbpoweramp software, created a new folder and transferred both to my NAS (D-link 323 and Naim approved). What happened is that although the folders are recognized as two different ones on the NAS, the HDX software sees this as one artist, the second folder was never recognized by the HDX IPUI. Therefore,when I selected Peter Bernstein Trio and "added to current playlist" all 24 tracks were there and listed as 1-12/13-24. For some reason my album art did not load so according to the HDX IPUI I could not differentiate between the file types. In essence, I unknowingly created a double blind study.
Listed to the first track "Let's Cool One" and then the same track again, but with the second unknown file type.
Within the first 10 seconds there was a big difference between the recordings. The second had a lower tone, more dynamics, presence and feel. Going back again to the first it was louder, muddled and missing the dynamics of the second track.
I repeated the "experiment" with the second track "Pannonica"-- same result. The same for all tracks.
I then opened up the DTC under now playing which listed the tracks in the same order as my "Current Playlist", but this time I got to see what the file types were:
Tracks 1-12-- FLAC
Tracks 13-24-- WAV
Now I know the argument about bit perfect, etc... I am also aware that FLAC contains the same material as a WAV file, just encoded differently.
However, the ears say differently. I cannot compare with anyone else's system or setup, just within the confines of mine.
The difference were there, no question and they were not subtle.
I downloaded a 24/96 album from HDTracks by Peter Berstein called Monk. Really good album BTW, his take from a guitar trio perspective on the music of TS Monk.
After downloading the album, I then converted it to WAV using Dbpoweramp software, created a new folder and transferred both to my NAS (D-link 323 and Naim approved). What happened is that although the folders are recognized as two different ones on the NAS, the HDX software sees this as one artist, the second folder was never recognized by the HDX IPUI. Therefore,when I selected Peter Bernstein Trio and "added to current playlist" all 24 tracks were there and listed as 1-12/13-24. For some reason my album art did not load so according to the HDX IPUI I could not differentiate between the file types. In essence, I unknowingly created a double blind study.
Listed to the first track "Let's Cool One" and then the same track again, but with the second unknown file type.
Within the first 10 seconds there was a big difference between the recordings. The second had a lower tone, more dynamics, presence and feel. Going back again to the first it was louder, muddled and missing the dynamics of the second track.
I repeated the "experiment" with the second track "Pannonica"-- same result. The same for all tracks.
I then opened up the DTC under now playing which listed the tracks in the same order as my "Current Playlist", but this time I got to see what the file types were:
Tracks 1-12-- FLAC
Tracks 13-24-- WAV
Now I know the argument about bit perfect, etc... I am also aware that FLAC contains the same material as a WAV file, just encoded differently.
However, the ears say differently. I cannot compare with anyone else's system or setup, just within the confines of mine.
The difference were there, no question and they were not subtle.
Posted on: 07 March 2010 by JamH
Have you tried converting WAV to FLAC [WAV-1] and then converting the FLAC back to WAV [WAV-2] and comparing the two WAV files ?
James H
James H
Posted on: 07 March 2010 by js
This was effectively already done as the files were at some point wav before being reduced to flac for download. I don't think there's any question here about maintaining content, simply perceived differences in playback.
Posted on: 07 March 2010 by luxen2
To me, wav sound better.
Someone told me that the "on-the-spot" "transformation from flac to wav makes it sound a little worse...
My ears like wav more.
Someone told me that the "on-the-spot" "transformation from flac to wav makes it sound a little worse...
My ears like wav more.
Posted on: 07 March 2010 by JamH
All this is worrying ..
I am using a non-Naim source into a Naim amplifier [82/2xHiCap+250] ...
I started ripping to WAV with EAC [Exact Audio Copy] but am now ripping to FLAC with dbPoweramp.
I suppose I could just convert the FLAC back to WAV.
James
P.S. Source is Logitec Transporter into Resolution Audio Opus-21 as DAC.
I am using a non-Naim source into a Naim amplifier [82/2xHiCap+250] ...
I started ripping to WAV with EAC [Exact Audio Copy] but am now ripping to FLAC with dbPoweramp.
I suppose I could just convert the FLAC back to WAV.
James
P.S. Source is Logitec Transporter into Resolution Audio Opus-21 as DAC.
Posted on: 07 March 2010 by u5227470736789439
Not worrying at all!
Did you notice a drop in quality when using FLAV [from EAC], and if not, just carry on enjoying the results.
ATB from George
Did you notice a drop in quality when using FLAV [from EAC], and if not, just carry on enjoying the results.
ATB from George
Posted on: 07 March 2010 by pcstockton
quote:Originally posted by luxen2:
To me, wav sound better.
Someone told me that the "on-the-spot" "transformation from flac to wav makes it sound a little worse...
My ears like wav more.
Would you feel differently if someone had told you FLAC sounded better?
Posted on: 07 March 2010 by JamH
Dear George.
To be honest I am hardly listenning to my hi-fi .. I am spending all my time 'ripping' ....
The main aim is to tidy up the house [get stuff off the floor] and get stuff organised ..
A side effect of 'ripping' is that I have always worried about the house burning down and destroying my CD collection -- now I can store it all on a hard-drive and get a friend to store a copy.
But -- back to the original question -- I don't want to use the wrong format and then find when I get a better hi-fi that I made a mistake [like thinking MP3 is OK until you have a good system].
James
To be honest I am hardly listenning to my hi-fi .. I am spending all my time 'ripping' ....
The main aim is to tidy up the house [get stuff off the floor] and get stuff organised ..
A side effect of 'ripping' is that I have always worried about the house burning down and destroying my CD collection -- now I can store it all on a hard-drive and get a friend to store a copy.
But -- back to the original question -- I don't want to use the wrong format and then find when I get a better hi-fi that I made a mistake [like thinking MP3 is OK until you have a good system].
James
Posted on: 07 March 2010 by u5227470736789439
Both FLAC and ALAC are easily returned to WAV status, if you ever think you can hear a difference. There is reason that they are called lossless formats. I have yet to see a real argument that something is lost in the process, [one way or the other to and from ALAC or FLAC to and from WAV,] with competent processing.
If ALAC were compromised compared to WAV then CD replay would beat my iTunes arrangement decisively, which till I compare with CD replay costing many thousands so far it has not. Even then the best CD replay does not leave it feeling very different or necessarily less fine. Possibly there is a slight presentaional difference here and there, but this will be a matter of individual taste and arguable as to which is actaully better in one person's opinion or another's. Auditioning might help in the issue. There is no decisive winner, IMHO.
I would stop worrying, as you rips are not compromised in any case.
ATB from George
If ALAC were compromised compared to WAV then CD replay would beat my iTunes arrangement decisively, which till I compare with CD replay costing many thousands so far it has not. Even then the best CD replay does not leave it feeling very different or necessarily less fine. Possibly there is a slight presentaional difference here and there, but this will be a matter of individual taste and arguable as to which is actaully better in one person's opinion or another's. Auditioning might help in the issue. There is no decisive winner, IMHO.
I would stop worrying, as you rips are not compromised in any case.
ATB from George
Posted on: 07 March 2010 by JamH
George,
Really I should rip to WAV [WAV-1] then to FLAC ..
Then convert WAV to FLAC [WAV-2] and compare the two WAV versions ..
That would reassure me that at least if I was using the wrong format I culd fix things ..
James
Really I should rip to WAV [WAV-1] then to FLAC ..
Then convert WAV to FLAC [WAV-2] and compare the two WAV versions ..
That would reassure me that at least if I was using the wrong format I culd fix things ..
James
Posted on: 07 March 2010 by u5227470736789439
It would be reassuring, and that is what matters. Tagging WAV files is a nightmare if you go that way!
ATB from George
ATB from George
Posted on: 07 March 2010 by JamH
Agreed ..
So back to the original question : does WAV sound better than FLAC
James
So back to the original question : does WAV sound better than FLAC
James
Posted on: 07 March 2010 by u5227470736789439
If the recreation of the digital signal from the file held on the Hard Drive, that is sent to the DAC, is competent then a lossless file is just that.
As I said about ALAC to WAV, it might be possible to believe one might be able to detect a tiny difference, but blind I am sure that the great majority of people would not be able to tell which was which!
In a word, "No," but others will no doubt claim different.
James, you are a music enthusiast. I know that.
Please do the experiements, satisfy yourself, and get back to what matters ... Well that would be my advice!
ATB from George
As I said about ALAC to WAV, it might be possible to believe one might be able to detect a tiny difference, but blind I am sure that the great majority of people would not be able to tell which was which!
In a word, "No," but others will no doubt claim different.
James, you are a music enthusiast. I know that.
Please do the experiements, satisfy yourself, and get back to what matters ... Well that would be my advice!
ATB from George
Posted on: 07 March 2010 by David Dever
quote:Originally posted by gary1 (US):
OK, here we go.
I downloaded a 24/96 album from HDTracks by Peter Berstein called Monk. Really good album BTW, his take from a guitar trio perspective on the music of TS Monk.
After downloading the album, I then converted it to WAV using Dbpoweramp software, created a new folder and transferred both to my NAS (D-link 323 and Naim approved). What happened is that although the folders are recognized as two different ones on the NAS, the HDX software sees this as one artist, the second folder was never recognized by the HDX IPUI. Therefore,when I selected Peter Bernstein Trio and "added to current playlist" all 24 tracks were there and listed as 1-12/13-24. For some reason my album art did not load so according to the HDX IPUI I could not differentiate between the file types. In essence, I unknowingly created a double blind study.
Listed to the first track "Let's Cool One" and then the same track again, but with the second unknown file type.
Within the first 10 seconds there was a big difference between the recordings. The second had a lower tone, more dynamics, presence and feel. Going back again to the first it was louder, muddled and missing the dynamics of the second track.
I repeated the "experiment" with the second track "Pannonica"-- same result. The same for all tracks.
I then opened up the DTC under now playing which listed the tracks in the same order as my "Current Playlist", but this time I got to see what the file types were:
Tracks 1-12-- FLAC
Tracks 13-24-- WAV
Now I know the argument about bit perfect, etc... I am also aware that FLAC contains the same material as a WAV file, just encoded differently.
However, the ears say differently. I cannot compare with anyone else's system or setup, just within the confines of mine.
The difference were there, no question and they were not subtle.
Were you running 1.5a software on the HDX? If not, you might have an older version of the FLAC decoder, which WILL affect your results.
Posted on: 08 March 2010 by gary1 (US)
Nope, waiting for the update.
Will have to ry again afterwards. I think JS has the new version, not sure, and there was still a difference-- we will see.
Will have to ry again afterwards. I think JS has the new version, not sure, and there was still a difference-- we will see.
Posted on: 08 March 2010 by js
Just some of the beta installed so probably not. As when you called with your personal findings, I really don't care for the fact that FLAC has been quite audible on everything I've tried it on. I really would prefer lossess to be as good in active use. I'll get the new software up and see if Naim have sorted it out. My guess is that they've given it extra attention now that flac is a rip option.
Posted on: 08 March 2010 by Hans Vereecken
Were you running 1.5a software on the HDX? If not, you might have an older version of the FLAC decoder, which WILL affect your results.[/QUOTE]
Does it matter in which version the WAV has been compressed to FLAC. I'm now using the latest FLAC encoder in DBPowerAmp but in the past I used the older versions.
Does it matter in which version the WAV has been compressed to FLAC. I'm now using the latest FLAC encoder in DBPowerAmp but in the past I used the older versions.
Posted on: 08 March 2010 by {OdS}
quote:Originally posted by gary1 (US):
The difference were there, no question and they were not subtle.
Gary,
To me this seems like a software "issue". I have no doubt that Flac and Wave sound exactly the same from a bit per bit point of view, if that means anything at all. Now, dbPowerAmp might be very good at playing/reading/whatever Flac and migth therefore create a better sounding (at least to your ears) Wave file than the source Flac file played by another software.
Christian
Posted on: 08 March 2010 by js
quote:Originally posted by Hans Vereecken:
Were you running 1.5a software on the HDX? If not, you might have an older version of the FLAC decoder, which WILL affect your results.
Does it matter in which version the WAV has been compressed to FLAC. I'm now using the latest FLAC encoder in DBPowerAmp but in the past I used the older versions.[/QUOTE]Encoder shouldn't really matter but I would use '0' setting on encode to keep decoding as easy as possible.
Posted on: 08 March 2010 by DaveBk
A few thoughts:
Are the perceived differences between WAV and FLAC only noticed on 'single processor' platforms - by this I mean devices where the decode of the FLAC stream is performed by the same processor as the one that's feeding the resulting PCM data to the DAC chip? I wonder if the decode of each FLAC block could be taking slightly different amounts of time and thus introducing some jitter into the steam of data being passed to the DAC chip?
I started thinking about how FLAC works - analyse the audio data in a block, find a function that approximates the data, subtract this from the original, encode the difference as efficiently as possible. Depending on the approximation function each block could take a different number of instructions to decode, so to feed the data into the DAC at a precise and constant rate the decode process would need to buffer the data and clock the data into DAC using a realtime hardware interrupt. Any error or lack of precision would impact the precise time that the next sample is latched onto the DAC chip.
I'm wondering if this could explain the perceived differences....
In my setup, the Transporter is doing the decode and sending the resulting PCM stream to the Naim DAC. Any 'Decode Jitter' would be isolated to to Transporter end of the SPDIF cable. The Sharc DSP's in the DAC have to buffer the data and then apply the interpolation filters before sending the data to the DAC chip, but if this process is more linear in the time domain than the FLAC decode it would introduce less processing jitter?
I have not managed to do any listening tests yet, but am very happy with how my system sounds... My analytical brain wants to find a theoretical basis to the perceived differences that some people report between WAV and FLAC - anyone care to comment on my thoughts above?
Thanks, Dave.
Are the perceived differences between WAV and FLAC only noticed on 'single processor' platforms - by this I mean devices where the decode of the FLAC stream is performed by the same processor as the one that's feeding the resulting PCM data to the DAC chip? I wonder if the decode of each FLAC block could be taking slightly different amounts of time and thus introducing some jitter into the steam of data being passed to the DAC chip?
I started thinking about how FLAC works - analyse the audio data in a block, find a function that approximates the data, subtract this from the original, encode the difference as efficiently as possible. Depending on the approximation function each block could take a different number of instructions to decode, so to feed the data into the DAC at a precise and constant rate the decode process would need to buffer the data and clock the data into DAC using a realtime hardware interrupt. Any error or lack of precision would impact the precise time that the next sample is latched onto the DAC chip.
I'm wondering if this could explain the perceived differences....
In my setup, the Transporter is doing the decode and sending the resulting PCM stream to the Naim DAC. Any 'Decode Jitter' would be isolated to to Transporter end of the SPDIF cable. The Sharc DSP's in the DAC have to buffer the data and then apply the interpolation filters before sending the data to the DAC chip, but if this process is more linear in the time domain than the FLAC decode it would introduce less processing jitter?
I have not managed to do any listening tests yet, but am very happy with how my system sounds... My analytical brain wants to find a theoretical basis to the perceived differences that some people report between WAV and FLAC - anyone care to comment on my thoughts above?
Thanks, Dave.
Posted on: 09 March 2010 by luxen2
quote:Originally posted by pcstockton:quote:Originally posted by luxen2:
To me, wav sound better.
Someone told me that the "on-the-spot" "transformation from flac to wav makes it sound a little worse...
My ears like wav more.
Would you feel differently if someone had told you FLAC sounded better?
Nope.
I checked it out many times on different systems (PC and HiFi).
WAV-files sound better to me.
Posted on: 09 March 2010 by likesmusic
quote:Originally posted by js:
Encoder shouldn't really matter but I would use '0' setting on encode to keep decoding as easy as possible.
It isn't necessarily the case that FLAC decoding compression level '0' is 'easier' than, say, level 5. Setting a higher compression level allows the encoder to spend more time finding a potentially more efficient strategy, which may be quicker to decode. After all , the more compression the less data there is for the decoder to process. According to FLAC's own tests here Level 5 offers the least decode time. But the results vary according to the type of music.
Whichever level you use though, if the results sound different from the equivalent WAV file, then something is broken somewhere, and I very much doubt it is the mathematics of FLAC.
Posted on: 09 March 2010 by pcstockton
Likes,
I have been using level 5 forever. It's nice to simply trust people who have experience. When i set-up EAC intitally i recall something to that effect about 5 being most economical all the way around.
I have been using level 5 forever. It's nice to simply trust people who have experience. When i set-up EAC intitally i recall something to that effect about 5 being most economical all the way around.
Posted on: 09 March 2010 by likesmusic
pcstockton - i use level 5 because i can't really be bothered adjusting it from the dbPoweramp default. If you checkout FLACs own data here you can see that the differences in the amount of space saved or the decoding speed aren't enormous between compression levels, but the differences in the time taken to encode vary significantly, so it seems that 5 is a decent compromise. As I understand it, all the hard work is done on the encoding side; decoding is meant to be fast and simple; the differences in decoding time between level 0 and 8 are small, and don't correlate with the level, so it seems to me potentially misguided for js to rip at level 0 in the belief that this makes things easiest for the decoder. And in any case, if I had a bit of kit that made FLAC (-8) sound different to FLAC (-1) I'd take it back. It all varies with the type of music anyway. I wonder whether Mozart compresses more than Stockhausen?
Posted on: 09 March 2010 by u5227470736789439
I wonder whether Mozart compresses more than Stockhausen?
If you compress Stockhausen enough does it become as diamond-like in its perfection as Mozart already is without any compression at all?
Now there's a question!
ATB from George
If you compress Stockhausen enough does it become as diamond-like in its perfection as Mozart already is without any compression at all?
Now there's a question!
ATB from George
Posted on: 09 March 2010 by PureHifi
quote:Originally posted by gary1 (US):
After downloading the album, I then converted it to WAV using Dbpoweramp software
Gary, can I just ask if you maintained the higher 24bit rate in the WAV encoding or did you convert a 24/96 FLAC file to a 16/44.1 WAV file ?