View Full Version : In the process of converting Flac files to Wav..have I gone mad!? Well, no actually..
Been thinking about this for a while, doing research on User opinion on various forums and MORE IMPORTANTLY listening with my own ears. :D
I started a thread over here:
http://yabb.jriver.com/interact/index.php?topic=75097.0
The usual snag with WAV is that you can lose Tagging information, however this is not a problem with Jriver.
Yep, I know that when a Flac is uncompressed IT IS THE SAME BIT FOR BIT as the original Wav file, no argument here. :)
After spending a lot og time experimenting with this stuff, my opinion is that its the "on the fly" conversion of Flac to PCM before it is sent to your Dac that makes the difference.
Its an extra process in the chain before the music file is sent to your Dac.
Why have it if you don't need to?
When I ran a Squeezebox Touch one of the more effective tweaks was to have Flac decoding done by the Server as opposed to the Touch itself.
I think this is a similar principle to what is happening here, albeit not due to hardware limitations as was the case with the Touch.
No, I think its the extra software decoding process that's doing something.
Whether its the Flac codec itself that's not 100% or its a Windows/CPU cycles phenomenon I don't know...
and having spent hours comparing with all sorts of different types of music I don't care!
there IS a difference - not night and day - but you miss it when its not there.
This is just my view, I'm not out to argue my case or justify what I think, just putting it out there for others to try or ignore.
Either way is cool.
So thats my strategy and I don't need to think about it anymore.
Fit(convert) 'n forget.
Use Wav for playback, use Flac for my backup.
Try converting a flac of some music you know really well to Wav..see what you think.
Your findings are similar to mine, and i fully agree with your statement that "something is missing" when playing losslessly compressed files, whether they are ALAC or FLAC doesn't seem to matter.
WAV and AIFF have some "air" and clearer definition of what going on in the music.
Regards Mike
I've done a few experiments with FLAC, a SBtouch, LMS, and my M-DAC's 'bit perfect' test...
http://theartofsound.net/forum/showthread.php?t=21132
...from a data supply perspective one can't do better than 'bit perfect', into the DAC via asynchronous USB (with a galvanic isolator).
However there may be other factors affecting you sound quality Gaz, if your touch is transcoding (ie. working more), and you're not galvanically isolated from your DAC (are you using coax?) then perhaps that's what's degrading your sound...i.e. the secondary effects of transcoding, not the data itself, particularly if you're using a synchronous connection...Jitter? Interference etc.?
then perhaps that's what's degrading your sound...i.e. the secondary effects of transcoding, not the data itself.
Yes, I believe that's what he's saying.
I use JPLAYmini, there's another reason for me to use WAV - JPLAY converts flac to wav and stores both in memory. This in-memory player sounds best with smaller memory allocations so flac in this case just wastes space and degrades sound quality due to increased memory requirements.
Yes, I believe that's what he's saying.
I use JPLAYmini, there's another reason for me to use WAV - JPLAY converts flac to wav and stores both in memory. This in-memory player sounds best with smaller memory allocations so flac in this case just wastes space and degrades sound quality due to increased memory requirements.
So it's rather dependant on your DAC's architecture, and connection type...:eyebrows:
As I'm using an asynchronous USB connection I should be jitter free, and as I'm using a USB isolator, I should be interference free, and as I'm getting 'Bit Perfect' at the DAC I can't see where the improvement will come from :eyebrows:
However in the interest of science I'll have a little experiment when I get some time :)
Hi Gaz, yep, I agree with WAV / AIFF sounding marginally better than FLAC / ALAC. As you may have read on another thread I verified this via Foobar2000's ABX comparator, so the naysayers couldn't hit back at me with the "yeah, but did you do a double blind test?" line... ;)
So it's rather dependant on your DAC's architecture, and connection type...:eyebrows:
As I'm using an asynchronous USB connection I should be jitter free, and as I'm using a USB isolator, I should be interference free, and as I'm getting 'Bit Perfect' at the DAC I can't see where the improvement will come from :eyebrows:
However in the interest of science I'll have a little experiment when I get some time :)
It does depend on the details of the setup. I too use async but bit perfect can be a long way from jitter free (bit perfect just means you have all the bits present and in the right order, it says nothing about timing). It seems that most dacs do very little internal buffering (even though they often claim differently). So they are very dependent on the timing of the data arriving at the dac input. Breaking the data down into the smallest packets possible for a given PC seems IME gives the best quality. This is what JPLAY does, it does its level best to turn windows into an RT O/S so the data is properly timed.
So it's rather dependant on your DAC's architecture, and connection type...:eyebrows:
As I'm using an asynchronous USB connection I should be jitter free, and as I'm using a USB isolator, I should be interference free, and as I'm getting 'Bit Perfect' at the DAC I can't see where the improvement will come from :eyebrows:
However in the interest of science I'll have a little experiment when I get some time :)
My setup:
Server >asynchronous usb> Vlink192 >coax> Rega Dac
The Vlink has galvanic isolation as does the Rega.
(dont use a Touch)
There is an additional process in the chain when Flac is uncompressed before hitting your dac which you don't get playing from Wav...and to me this is where the difference must come from.
I agree bitperfect is bitperfect, not suggesting otherwise. :eyebrows:
*Edit*
Converting to AIFF rather than WAV
wee tee cee
31-10-2012, 17:48
Loaded j river a few days ago....I have wav and i tunes losless tracks in duplicate for a few tracks. I was playing around last night and noticed a slight but noticeable difference between the two but thought it was just my imagination.....really should trust my ears more....
Ammonite Audio
31-10-2012, 18:25
My consistent experience is that WAV results in more musically satisfying reproduction than FLAC, particularly when I used Foobar2000 - the difference between FLAC and WAV narrowed when I moved to J River, but is still there. How much of the difference is simply due to different FLAC codecs and inefficient conversion? How can we be sure that we are comparing eggs with eggs, as it were? I can't be arsed with any of that, so I use WAV almost exclusively. J River manages to tag WAV files, so there are no problems with file organisation.
Lets reiterate whats been said up thread it's not the file format itself but the on-the-fly transcoding that seems to bring about the apparent audible difference some peopple detect. Personally I've not noticed it and IVe done the Foobar ABX thing. What mechanism is at play here is anybody's guess.
On the file type itself, If you take a Wav file, create a duplicate, encode the duplicate, manually decode the Flac back to Wav and compare the duplicate to the original you will not find a difference. Perform a binary compare between the two and they will be identical.
It seems that most dacs do very little internal buffering (even though they often claim differently).
I'm not sure what size the M-DAC's buffer is, however it's very obvious that the DAC is drawing it's data source from there, and there alone...so asynchronous, and subject to it's own timing constraints, which should theoretically remove jitter.
It's quite easy to see that the buffer is the soul source, there is a buffer level meter. Whenever there is a dropout in the buffer there is a dropout in the sound. If you try to use an SBTouch with an M-DAC without a USB hub you'll get plenty of these :eyebrows:
With a hub it's solid at 50% :)
Stratmangler
01-11-2012, 10:09
I can't be arsed with all this mucking about with formats lark - I just play music these days :eyebrows:
I'm not sure what size the M-DAC's buffer is, however it's very obvious that the DAC is drawing it's data source from there, and there alone...so asynchronous, and subject to it's own timing constraints, which should theoretically remove jitter.
It's quite easy to see that the buffer is the soul source, there is a buffer level meter. Whenever there is a dropout in the buffer there is a dropout in the sound. If you try to use an SBTouch with an M-DAC without a USB hub you'll get plenty of these :eyebrows:
With a hub it's solid at 50% :)
I can't speak for the M-DAC as I've not played with one. What I do know it that JPLAY users report gains in sound quality due to reductions in buffers on the PC side even with async DACs using their own internal buffers. It's one of those things that is strange; you'd think that once the data is lined up in a DAC's buffer the manner in which it arrived should not matter, therefore Foobar, Jriver, JPLAY etc should all sound 100% identical if setup bitperfect. But they don't.
I did some testing with a squeezebox touch and server transcoding from FLAC to WAV a while back. I found that transcoding FLAC files to WAVs on the server rather than the Squeezebox Touch it increased CPU utilisation of the SBT about 25%. I presume this is the overhead of managing the network interface and buffer management within the Touch.
I thought that this was interesting as the rationale of using the touch toolbox 3.0 etc was to eliminate extra processing and allow the CPU to idle more, but this theory didn't appear to be bourne out in practice.
I couldn't tell the difference in a randomised play of WAV and FLAC versions of the same file either at all, despite listening hard for the increased openness etc. It sounds great either way to me, so I am pleased not to be suffering from format anxiety any more - am sticking with FLAC.
Cheers,
Alex
Stratmangler
01-11-2012, 10:20
I couldn't tell the difference in a randomised play of WAV and FLAC versions of the same file either at all, despite listening hard for the increased openness etc. It sounds great either way to me, so I am pleased not to be suffering from format anxiety any more - am sticking with FLAC.
Cheers,
Alex
It's a good place to be :)
Agreed!
Always happy to find cheap (i.e. free) things that can add to the enjoyment of my music.
Its something I'll do once and forget about, its all automated in Jriver so not a drama.
No anxiety just curiosity. :)
Gary,
Just out of curiosity, have you ever tried or noticed this difference using SPDIF, or do you connect to the MDAC via USB only?. I am wondering if this is a PC-specific difference, and whether the Windows software/driver stack is affected by FLAC decoding and USB processing. I know you said that you said you saw an improvement on the SBT too, but maybe it is DAC specific (I presume you were using a DAC with the SBT - which one?).
I have to say that in my system at least, I am not yet totally convinced that async USB is the panacea that it is cracked up to be. In fact, I am finding SPDIF to sound significantly better than USB with the SB Touch and my Cambridge Azur 851C. I have yet to try this via my laptop, but I think I may revisit this.
Cheers,
Alex
Gaz - do you use "Play Files From Memory Instead of Disk" in Jriver Media Center?
I do - and I can't detect a difference between FLAC and WAV. My theory is that it unpacks the FLAC file to WAV before buffering in memory so by chosing this, so you are effectively playing WAV anyway... As I said MY theory, and not sure if it is how it works or not.
I'm certainly not going to convert nearly 2TB of FLACs back to WAV though, as I just can't tell the difference.
Gary,
Just out of curiosity, have you ever tried or noticed this difference using SPDIF, or do you connect to the MDAC via USB only?. I am wondering if this is a PC-specific difference, and whether the Windows software/driver stack is affected by FLAC decoding and USB processing. I know you said that you said you saw an improvement on the SBT too, but maybe it is DAC specific (I presume you were using a DAC with the SBT - which one?).
I have to say that in my system at least, I am not yet totally convinced that async USB is the panacea that it is cracked up to be. In fact, I am finding SPDIF to sound significantly better than USB with the SB Touch and my Cambridge Azur 851C. I have yet to try this via my laptop, but I think I may revisit this.
Cheers,
Alex
Hi Alex,
I take USB out from my server into a Vlink 192.
Its coax spdif from there into my Rega dac.
The Rega doesn't do 24/96 over usb hence the Vlink converter.
USB direct from the server doesn't sound as good as usb/spdif converted from the Vlink.
I bypass the Windows audio subsystem completely by outputting audio with Jriver set to "Wasapi Event Style exclusive access" output.
This enables the bitstream to pass to the Dac unaltered.
With the Touch I felt the improvement came from minimising the strain on its limited resources by getting the server to do the Flac>PCM number crunching.
The difference in the PC sound is similar but comes more from (imo) the additional on-the-fly processing rather than limited resources.
What I feel I have maybe done is take an unnecessary step out the chain?
Gaz - do you use "Play Files From Memory Instead of Disk" in Jriver Media Center?
I do - and I can't detect a difference between FLAC and WAV. My theory is that it unpacks the FLAC file to WAV before buffering in memory so by chosing this, so you are effectively playing WAV anyway... As I said MY theory, and not sure if it is how it works or not.
I'm certainly not going to convert nearly 2TB of FLACs back to WAV though, as I just can't tell the difference.
hi Alex,
I dont use memory play with Jriver as I don't hear any improvement with it since going to v18 of Jriver.
Also, if your pulling in your Flacs over a network it requires copying the file across the network INTO MEMORY ANYWAY before any playback can occur.
So having the "memory play" button ticked is redundant as files are loaded into memory anyway if your playing over a network.
I had a conversation with Jim Hillegass (the Jriver guy) about this and he confirmed it.
Good discussion about it here:
http://yabb.jriver.com/interact/index.php?topic=75097.0
and here
http://yabb.jriver.com/interact/index.php?topic=75123.0
The last thing I wanted to do was to convert ANYTHING (I have a similar sized library to you Alex and still relive the horror of ripping all those CDs :steam:) and I looked in to different ways to achieve uncompressed audio from Jriver without converting.
I tried a thing called AssetUPnP, which is an audio only DNLA server?
The Naim and Linn guys talk a lot of this in their forums as it does the conversion process first THEN sends the uncompressed PCM to Jriver for playback.
I tried it, but found it to be a bit fiddly and unreliable.
It wasn't finding all my music library and didn't always play nice with Jriver.
Playlists etc didn't really work the way they were supposed to and it would have meant a lot of work to recreate them in a way that did.
So in the interests of a simple life I decided against it.
Also, it was an extra process in the chain again. :scratch:
With Jriver you can select which library you listen from and I sat down and listened to the same tracks, swapping between the DNLA served "uncompressed" version and physical converted AIFF files on my HDD.
Playback from DNLA server was missing something that direct playback had.
They sounded subtly different and I preferred the direct playback.
Maybe down to network issues, who knows?
Given that it was a "do once and forget" thing with no fannying about involved and it just...worked...I decided to convert my library.
If you can get a 5% improvement (say) for nothing but your time then that works for me.
Subtle, but you miss it when its not there.
All of this stuff is based on what I'm hearing with my ears in my system in my room by the way :)
Thanks Gaz.
Interestingly, I find that I prefer the sound directly from an external USB HDD, so I thought that might be the difference, but this would disprove either theory:
Vincent Cars Said:
I’m afraid memory playback as implemented in JRiver won’t help us.
It is an input buffer. The track is read and stored in memory.
The decoding is done during playback.
Memory playback as implemented in Foobar/Cplay/ SignalistHQ read the track and do the decoding on the fly. What is stored in memory is an output buffer with “DAC ready” output.
This eliminates both the differences in I/O and processing regardless of the format.
This is the trick to eliminate any possible audible difference between (lossless) formats.
So I can see why it is possible that the decoding of a FLAC file might interfere with other processes - and therefore sound quality...
Luckily, as I can't notice any difference I shall carry on in blissful denial with my FLAC library! ;)
ok...
so that's the "B's" done...24 more to go :lol:
I've read the articles in Stereophile and Absolute Sound and others who mentioned that converting the FLAC to WAV makes an improvement. All of my files are now in WAV format, attached to a 500-GB external HDD via USB to my Logitech SB Touch.
I have downloaded a bunch of 16/44 and 24/96 from HDtracks.com in FLAC format, and converted them to WAV. I have in many cases left the FLAC and WAV copies of the same files in the folder so I can listen to the FLAC version of a song, and then skip to the WAV of the same piece, and so on.
In every case, I have much preferred the WAV over the FLAC, and for me it was not even really close. The FLAC to my ears have sounded a bit muddy and congested compared to the WAV which have more definition and air and quieter background. If you have not convered FLAC to WAV yet, give it a try and trust your ears. You might not go back.
morris_minor
03-11-2012, 11:20
I'm quite happy myself with FLACs but I was wondering how folk who stream WAVs tag them?
I'm quite happy myself with FLACs but I was wondering how folk who stream WAVs tag them?
The same way I tag my LPs :) Actually I use a mix of FLAC and WAV but I can't be bothered with the UI as I use a player aimed at the best sound and minimising the UI is part of achieving this. If the UI with great tagging is important then you MAY need to compromise. If you find in-memory players don't increase sonics then you can happily use FLACs and tag them to your preferences.
Stratmangler
03-11-2012, 11:31
ok...
so that's the "B's" done...24 more to go :lol:
You are retaining the FLAC library aren't you, just in case something comes along sometime down the line which makes the FLAC stuff sound better?
That way you'd still keep all the tagging too :)
morris_minor
03-11-2012, 11:39
The same way I tag my LPs :) Actually I use a mix of FLAC and WAV but I can't be bothered with the UI as I use a player aimed at the best sound and minimising the UI is part of achieving this. If the UI with great tagging is important then you MAY need to compromise. If you find in-memory players don't increase sonics then you can happily use FLACs and tag them to your preferences.
Ok. Then I really am happy to stick with FLACs for all my Squeezeboxes!:D
Stratmangler
03-11-2012, 12:26
Gary
You're not converting all of your music files a folder at a time are you?
I think I'd set up Foobar to batch convert the lot in one hit :eyebrows:
I got a few files on FLAC and just tried this on some Ben Harper and will be doing this with the remaining files
If you (general) had followed my guide and ripped a WAV copy as reference in the first place you wouldn't be pissing about with batch convertors and re-rips now. :ner:;)
Also Stan mentioned this in the past and many thought he was a bit crazy
With Jriver, you can convert your Flacs to Wav without losing any tag information.:)
Also, I found AIFF format (Apples's version of Wav) to be slightly better than Wav.
As I said in the Title of this thread...have I gone mad? :stalks:
Woulld of thought they are the same code?
brian2957
04-11-2012, 08:30
With Jriver, you can convert your Flacs to Wav without losing any tag information.:)
Also, I found AIFF format (Apples's version of Wav) to be slightly better than Wav.
As I said in the Title of this thread...have I gone mad? :stalks:
What do you mean ' gone ' :ner:
bobbasrah
04-11-2012, 09:33
Woulld of thought they are the same code?
Think that is where we have been looking previously in error John, it may not be the file type which is the issue but the way the file is handled. ie - We attribute the change to the file not the environment.
When Stan raised this last in relation to WAV v FLAC, everything from filesize to filetype was alluded to and examined in a thread specific to it.
Whereas the difference could not be confirmed in the course of testing it out on everything from SB to computers on various platforms and settings, the number of variables in play on the equipment and player side were huge.
There is no doubt that differences exist, however small, so ALAC v FLAC v AIFF v WAV difference is entirely possible.
I can get the difference across different codes, just struggling to get my head around the difference in the same code....Sorry still a bit confussed by this
Stratmangler
04-11-2012, 10:01
As I said in the Title of this thread...have I gone mad? :stalks:
Possibly.
Still, if it makes you happy....:eyebrows:
Stratmangler
04-11-2012, 10:15
Think that is where we have been looking previously in error John, it may not be the file type which is the issue but the way the file is handled. ie - We attribute the change to the file not the environment.
When Stan raised this last in relation to WAV v FLAC, everything from filesize to filetype was alluded to and examined in a thread specific to it.
Whereas the difference could not be confirmed in the course of testing it out on everything from SB to computers on various platforms and settings, the number of variables in play on the equipment and player side were huge.
There is no doubt that differences exist, however small, so ALAC v FLAC v AIFF v WAV difference is entirely possible.
I agree.
I've also gone down the route of comparing fle types (having levelled the playing field playback wise) and concluded that I can't pick up any difference between playback on the four file types mentioned in the quoted post in my system in my home.
bobbasrah
04-11-2012, 10:20
I can get the difference across different codes, just struggling to get my head around the difference in the same code....Sorry still a bit confussed by this
Understandable John.... What I was trying to get at is that the decompression/player code and the electronic environment in which it runs may handle even similarly structured file types with subtle differences..... ;)
I spent the best part of a couple of day's testing formats, especially FLAC and WAV before commencing the task of ripping my entire collection (its been discussed before as John points out). I settled on FLAC as I could not discern any difference. I'm not disagreeing that there may be an audible difference, just that I personally can't hear one, so for me FLAC is where I am and where I'm staying. I am more than happy with what I have and much prefer to spend my time listening to music ;)
dave2010
05-11-2012, 08:48
Odd. We trod these paths a while back, and Stan asserted rather strongly IIRC that wav sounded better. There also seemed to be some who said that there was no difference, and quite a number who stated, and apparently verified, that bit perfect streams really work. If there is a difference, positive or otherwise, surely it must be due to secondary effects - timing, power etc. I queried the coding in the previous thread, but several said that the flac or alac coding had been well tested, so software problems were ruled out as a source of differences.
bobbasrah
05-11-2012, 09:43
Odd. We trod these paths a while back, and Stan asserted rather strongly IIRC that wav sounded better. There also seemed to be some who said that there was no difference, and quite a number who stated, and apparently verified, that bit perfect streams really work. If there is a difference, positive or otherwise, surely it must be due to secondary effects - timing, power etc. I queried the coding in the previous thread, but several said that the flac or alac coding had been well tested, so software problems were ruled out as a source of differences.
This is where we have perhaps been mis-directed Dave.
The motherboard components, OS, and critically player software coding, may account for the differences. The focus at the time of that FLAC/WAV thread looked to file characteristics etc as a logical reason for the difference, rather than the environment.
Vincent Kars
05-11-2012, 15:41
Many have done the test
WAV1 > FLAC/ALAC/AIFF > WAV2
All these test yield the same result: WAV1 and WAV2 are bit identical.
A testing method a bit rare is to record e.g. the SPDIF out. Now you capture exactly what has been send to the DAC.
This is a bit more complex but what I have found on the internet again is zero difference.
It is highly unlikely that the difference is at bit level.
Some time ago, there was some stir because of an article in TAS.
Guys did WAV1 > FLAC > WAV2 and where hearing a difference between WAV1 and WAV2.
Their conclusion: FLAC degrades the sound quality permanently!
Of course they were to ignorant to do a bit wise comparison between WAV1 and WAV2
This I do think a typical case of hearing differences simply due to our perception.
Our perception is horribly unreliable; it lures us in hearing differences that are not there.
A technical answer in my opinion is true memory playback.
If all I/O and all decoding is done before playback starts, we have a buffer with identical “DAC ready” output.
This should eliminate all possible differences due to software-induced jitter.
BTW: a nice example how I was fooled by my perception can be found here: http://www.whatsbestforum.com/showthread.php?7742-Me-and-my-perception
As I have been a recent convert to using a Linn Majik DS outputting to a Beresford Bushmaster (quite outstanding sound quality..and elevates the Majik to a far higher level) I have spent a fair amount of time of late ripping CD's and downloading Flac files..all stored on a QNAP twin drive NAS thingy, as well as a few on a PC laptop.
This topic was of obvious interest, and I was also vaguely aware of the subject of transcoding on the fly.
So half an hour ago I downloaded MinimServer onto my laptop as I had been reading that a recent update has given it the ability to play wav files from flac. Then Linn's Kinsky pics up the server for the DS to play.
Sure enough.. Just listening to Julian Bream and John Williams 'Together' as a wav file transcoded from the original flac sounds better. Individual guitars have a bit more air and definition around them.
Still have to figure out how to get MinimServer onto the NAS..but seems like the way to go.
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.