PDA

View Full Version : FLAC onboard decoding v. server side in SB2



Timbo
2005-07-10, 18:00
Hi there folks - I wonder if anyone can comment on my findings here as I think my brain has seized (well it is 1:30am and I shouldn’t be playing with my Squeezebox at this time of night...;-)

Anyway after reading all the advice on the forum I eventually settled (after much trial and error!) on EAC for ripping and FLAC for compression (I would prefer to use totally uncompressed WAV or AIFF as I have loads of space and a wired connection to the SB2 - but obviously no tags for WAV show through in SlimServer and as far as I can make out there is no native support in EAC for making AIFF files (unless I missed something?)

Anyway, just to check that FLAC really is no different to streaming the uncompressed WAV/AIFF file, I made a FLAC copy of an album (using external compression option in EAC) and a WAV copy (just clicking the WAV button in EAC) so I could compare the audio quality of each.

I cued the tracks in FLAC/WAV alternate order in SlimServer and went to have a listen. Instantly I played the first track and then it’s WAV counterpart it was obvious the WAV copy was better!

Now I have made sure SlimServer Player Settings/Audio/Bitrate Limiting is on ‘Unlimited’ (see I do read all the posts :-) - but I can tell the difference easily, no lengthy comparison required (although I did plenty of backwards and forwards testing on each track to make sure!) - the WAV file sounds more detailed within the first few seconds of listening.

Now as I see here posted (and on Hydrogenaudio) lots of times that ‘lossless’ means ‘lossless and no messing’ so I thought I better look into this a bit further. Obviously one of SB2’s new features is built-in hardware decoding of FLAC on the fly, so looking in Server Settings/File Types I came across lots of ‘convert this to that’ type tick box options - so - I un-ticked FLAC -> FLAC (built-in) and made sure FLAC -> WAV (flac) was ticked (WAV -> WAV ticked also of course).

Another set of listening tests later and now I really am confused, there might be a tiny difference, my ears are getting tired now, but it would appear to all intents and purposes that FLAC sounds pretty much the same as WAV...which of course it should really.

My conclusion is this (bearing in mind it is late and I might be hearing strange things!) - server side decoding of FLAC and then SlimServer sending the WAV down the line sounds better than SB2 decoding of FLAC on the fly.

Anyone else care to try this and let me know what they think?

PS. This SB2 is an amazing piece of kit - I have SB2/Chord 64 DAC feeding Meridian 502/557 into Ruark speakers and I am hearing things in the mix I didn’t with my Meridian 508.24 - missing a little subtlety and airiness perhaps but that could be the DAC being a bit forward - Meridian 566.24 DAC on it’s way to check this out :-)

sleepysurf
2005-07-10, 18:31
Well, I felt exactly the same way about audio quality comparing Wired vs. Wireless SB2. Finally conducted a BLIND listening test, which showed NO difference. See thread... http://forums.slimdevices.com/showthread.php?t=14811

Now, if you can score 70+% correct on a BLIND test, that would be significant.

kjg
2005-07-10, 21:57
Timbo wrote:

>Hi there folks - I wonder if anyone can comment on my findings here as I
>think my brain has seized (well it is 1:30am and I shouldn’t be playing
>with my Squeezebox at this time of night...;-)
>
>Anyway after reading all the advice on the forum I eventually settled
>(after much trial and error!) on EAC for ripping and FLAC for
>compression (I would prefer to use totally uncompressed WAV or AIFF as
>I have loads of space and a wired connection to the SB2 - but obviously
>no tags for WAV show through in SlimServer and as far as I can make out
>there is no native support in EAC for making AIFF files (unless I
>missed something?)
>
>Anyway, just to check that FLAC really is no different to streaming the
>uncompressed WAV/AIFF file, I made a FLAC copy of an album (using
>external compression option in EAC) and a WAV copy (just clicking the
>WAV button in EAC) so I could compare the audio quality of each.
>
>I cued the tracks in FLAC/WAV alternate order in SlimServer and went to
>have a listen. Instantly I played the first track and then it’s WAV
>counterpart it was obvious the WAV copy was better!
>
>Now I have made sure SlimServer Player Settings/Audio/Bitrate Limiting
>is on ‘Unlimited’ (see I do read all the posts :-) - but I can tell the
>difference easily, no lengthy comparison required (although I did plenty
>of backwards and forwards testing on each track to make sure!) - the WAV
>file sounds more detailed within the first few seconds of listening.
>
>Now as I see here posted (and on Hydrogenaudio) lots of times that
>‘lossless’ means ‘lossless and no messing’ so I thought I better look
>into this a bit further. Obviously one of SB2’s new features is
>built-in hardware decoding of FLAC on the fly, so looking in Server
>Settings/File Types I came across lots of ‘convert this to that’ type
>tick box options - so - I un-ticked FLAC -> FLAC (built-in) and made
>sure FLAC -> WAV (flac) was ticked (WAV -> WAV ticked also of
>course).
>
>Another set of listening tests later and now I really am confused,
>there might be a tiny difference, my ears are getting tired now, but it
>would appear to all intents and purposes that FLAC sounds pretty much
>the same as WAV...which of course it should really.
>
>My conclusion is this (bearing in mind it is late and I might be
>hearing strange things!) - server side decoding of FLAC and then
>SlimServer sending the WAV down the line sounds better than SB2
>decoding of FLAC on the fly.
>
>Anyone else care to try this and let me know what they think?
>
>PS. This SB2 is an amazing piece of kit - I have SB2/Chord 64 DAC
>feeding Meridian 502/557 into Ruark speakers and I am hearing things in
>the mix I didn’t with my Meridian 508.24 - missing a little subtlety and
>airiness perhaps but that could be the DAC being a bit forward -
>Meridian 566.24 DAC on it’s way to check this out :-)
>
>
>
>

Timbo,

I discovered this as well a while back and have since used PCM rather
than FLAC streaming with my unit. I find that at times the differences
are quite apparent, and at others not so much so. Its hard to draw
distinct conclusions from less than reproducible results. A more
definitive test might be to capture and record the PCM output of the
Squeezebox using both types of streaming and then compare the results. I
believe that there are PC sound cards that can do this (MAudio is one I
believe), but I unfortunately don't own one with such capabilities.

At the time I discovered this difference, I posted this result to the
newsgroup but was unable to provide anything but a subjective evaluation
so the thread quietly died out. However, it's good to know that others
can hear the same differences.

- Ken

seanadams
2005-07-10, 23:32
please back up these claims. You are saying that based on a non-blind subjective listen, our FLAC implementation is *broken*. The correctness of our FLAC decoder is empirical and all you need to do is save the bits at the output to test it.

Please note that our implementation:

1) is based on the official FLAC sources - it is the exact same code base as the one that's running on your computer.
2) has been tested by us for bit-perfect output, by recording PCM output back into a computer
3) has also been confirmed by us and others to pass through non-PCM bitstreams correctly
4) has also be tested by the author of FLAC, Josh Coalson, for compatibility with the included test suite

There are a couple known bugs and feature requests but none concerning accuracy, which is the entire point of using FLAC!

http://bugs.slimdevices.com/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&field0-0-0=product&type0-0-0=substring&value0-0-0=flac&field0-0-1=component&type0-0-1=substring&value0-0-1=flac&field0-0-2=short_desc&type0-0-2=substring&value0-0-2=flac&field0-0-3=status_whiteboard&type0-0-3=substring&value0-0-3=flac

Fabrice Rossi
2005-07-11, 09:21
Timbo a écrit :
> I cued the tracks in FLAC/WAV alternate order in SlimServer and went to
> have a listen. Instantly I played the first track and then it’s WAV
> counterpart it was obvious the WAV copy was better!

I have the same experience, but I managed to solve the problem with a
very simple trick (a lot of experiments were needed though). I use a
crystal wine glass (designed for Bordeaux wines, of course) half full of
sparkling water (San Pellegrino). With this put on top of my SB2, I can
hear any difference between flac files and wav files, whereas the
difference with night and day before. I also think the sound of both
files is way better than without. The improvement is clearer for flac
files, because they were really behind wav files.

Of course, you have to be careful because the top of the SB2 is not flat
so your glass can fall off. I use playdough (borrowed from my two year
old daughter) to build a small holding part. I tried several, but the
blue one clearly gives a much better sound, much more subtle than with
yellow one for instance. And keep away for cheap copies, you have to get
the original playdough.

As a French, I have of course listen to my system with wine instead of
sparkling water but I have to be honest, the best results come from the
San Pellegrino. Even Château Haut-Brion can't compete (neither Perrier,
what a shame for France). Also for the glass, you really need a Bordeaux
one. With a glass designed for Chimay beer (for instance), results are
really awful, especially in the bass.

Two last trick: you need bubbles in the water (so you have to change it
quite frequently) and the glass need a tuning period. It wont work out
of the box (and don't clean it to often).

Of course, I did not any blind test, because we all know this is BS.

Fabrice

Fifer
2005-07-11, 10:56
There's a good reason for this effect. If you read the label, you'll find that San Pellegrino is a mild diuretic and the perceived improvement is a result of taking the p**s.

Fabrice Rossi
2005-07-11, 11:31
Fifer a écrit :
> There's a good reason for this effect. If you read the label, you'll
> find that San Pellegrino is a mild diuretic and the perceived
> improvement is a result of taking the p**s.

Great explanation (as a poor French I dad to google a little bit to
really understand it...). But I thought that driking the wine would also
help in improving the sound, and it seems it does not. Being an
audiophile is really tough.

Fabrice

Timbo
2005-07-11, 11:36
please back up these claims. You are saying that based on a non-blind subjective listen, our FLAC implementation is *broken*. The correctness of our FLAC decoder is empirical and all you need to do is save the bits at the output to test it.

Please note that our implementation:

1) is based on the official FLAC sources - it is the exact same code base as the one that's running on your computer.
2) has been tested by us for bit-perfect output, by recording PCM output back into a computer
3) has also been confirmed by us and others to pass through non-PCM bitstreams correctly
4) has also be tested by the author of FLAC, Josh Coalson, for compatibility with the included test suite

There are a couple known bugs and feature requests but none concerning accuracy, which is the entire point of using FLAC!

http://bugs.slimdevices.com/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&field0-0-0=product&type0-0-0=substring&value0-0-0=flac&field0-0-1=component&type0-0-1=substring&value0-0-1=flac&field0-0-2=short_desc&type0-0-2=substring&value0-0-2=flac&field0-0-3=status_whiteboard&type0-0-3=substring&value0-0-3=flac

Hi Sean - well first of all am I correct in thinking that by deselecting the FLAC -> FLAC (built-in) and leaving FLAC -> WAV (flac) ticked that I am getting PC server side decoding anyway? I could be hearing anything making a difference if that isn’t the reason?

Secondly who said anything about your decoding implementation being *broken*, me-thinks you might be a tad overreacting here. I am really talking about the FLAC as compared to WAV issue - just thought the tick box made a difference that’s all - will test again in a moment.

Rest assured no criticism was intended of the most excellent SB2 - after all I have raved about it elsewhere and it is undeniably far too good for the price point (put one in a silver hi-fi box with half a brick to make it heavier and I would probably have paid 4 times the price - and still be amazed at the sound!! :-)

Anyway, sorry if my initial long post seemed griping, not intentional I assure you - you’re the technician and know about these things - I just listen to what my ears tell me and thought someone might be able to tell me why they seemed to be hearing what they were hearing?

OK - I am doing this as I write so this is fresh - I won’t get into built-in v server side decoding yet as I need to make sure I am hearing something real...so with all conversion tick boxes as they arrived first I compare FLAC to WAV.

I play a FLAC track (Kate Melua - Crawling up a Hill from Call of the Search) and it sounds great, nicely turned out and generally fine. I rewind and play the first few bars again and then immediately flick on to the WAV version that sits below in the playlist (I know, I know but I tried doing this blind and just kept ending up playing the wrong track!). The difference is harder to describe than I seemed to hear last night - there’s just slightly more, it’s slightly more impressive. The best way to describe it is after listening all the way through the FLAC track and feeling fine, then the WAV track starts to play and it’s as though your ears prick-up and take notice - I like it, I can’t rationalise exactly why but to me there just seems ‘more’. Ok so that shouldn’t be if bits are just bits and data is just data - but to my ears and in my system FLAC doesn’t do for me what WAV does.

Right - now let’s flick the switch on those conversions (leaving only FLAC -> WAV flac ticked in the FLAC options) and see what happens now...

OK this is more difficult, immediately I thought the FLAC track was the WAV track, then I checked and it wasn’t, then I rewound and listened to the first few bars again and flicked to the WAV track. Now I find it more difficult to tell the two apart. So I sit back and listen without trying to judge...and my foot is tapping to both tracks, maybe a slight edge to the WAV track but this is so slight now as to be easily dismissed as placebo...sorry.

OK that’s me done...will leave it to the rest of you guys to sort out why I am hearing what I think I am hearing - except to say maybe the SB2 decoding means the circuit is working harder and so the audio is a little less relaxed - I know it’s just digits..but..?

All I need to do now is find out how to tag WAV’s from EAC so I can see them on my SB2? Sorry to cause a stir :-)

Timbo
2005-07-11, 11:42
>Two last trick: you need bubbles in the water (so you have to change it quite frequently) and the glass need a tuning period. It wont work out of the box (and don't clean it to often).

You jest, but how true your words - the bubbles would no doubt oscilate at approx. 50-60khz and have an effect - however reflections from the glass itself would make measurement difficult....now maybe if we drink the wine instead of trying to place it on the SB2.....aah that sounds better already...now wheres my large spoon....;-)

seanadams
2005-07-11, 13:09
Secondly who said anything about your decoding implementation being *broken*, me-thinks you might be a tad overreacting here.

FLAC is by definition lossless. If you're not getting the same data out that you put in, then it's broken in every sense of the word.

Yannzola
2005-07-11, 14:49
I don't think Timbo is claiming that he isn't getting the same data out (bit wise)... only that he is hearing a difference between a raw WAV file sent directly to the SB vs. the same info sent as a FLAC file decoded onboard.

From my reading of Timbo's post, it seems he agrees that both the FLAC and the raw WAV file when sent as a WAV to the SB (bypassing the internal FLAC decoding) both sound essentially identical.

So... it seems to me the question not that the FLAC and WAV contain diffrent data, but how the process of onboard decoding potentially effects the output. Can the process involved in converting the FLAC onboard somehow effect the sound?

y.


FLAC is by definition lossless. If you're not getting the same data out that you put in, then it's broken in every sense of the word.

Timbo
2005-07-11, 15:27
FLAC is by definition lossless. If you're not getting the same data out that you put in, then it's broken in every sense of the word.
You said you tested your implementation (which after all is the same standard code in software regardless of where it is implemented) and you get out what you put in - so - it isn't broke.

Maybe what we measure is not all that is there? Or maybe what we do measure is not effecting the sound - maybe it is the actual act of deconstruction and reconstruction that effects something indefinable that we cannot measure - and perhaps when this is performed on a separate CPU a long way away from the audio circuits things sound better...?

Anyway, nuff of all this audiophile stuff - it doesn’t matter because your product is performing to specification and that specification is excellent!

What I would like to know from someone however is - bearing in mind I would really like to store my sound files uncompressed - then how do I get EAC to do this and not end up with a WAV file I can’t tag :-/

Timbo
2005-07-11, 15:37
I don't think Timbo is claiming that he isn't getting the same data out (bit wise)... only that he is hearing a difference between a raw WAV file sent directly to the SB vs. the same info sent as a FLAC file decoded onboard.y.

Exactly - thanks...:-)

seanadams
2005-07-11, 15:41
Can the process involved in converting the FLAC onboard somehow effect the sound?

Like I said in the wired vs wireless topic: I am not going to entertain the idea that there's a problem until someone can either HEAR or MEASURE it. Either of those tests is extremely easy to perform.

Yannzola
2005-07-11, 15:55
Okay,
I'm game. I'll try a HEARING test this eve... Sean, are the steps Timbo performed (deselecting all FLAC>XXX conversion options except for FLAC>WAV) the correct way to test for this?

y.


Like I said in the wired vs wireless topic: I am not going to entertain the idea that there's a problem until someone can either HEAR or MEASURE it. Either of those tests is extremely easy to perform.

seanadams
2005-07-11, 17:18
Okay,
I'm game. I'll try a HEARING test this eve... Sean, are the steps Timbo performed (deselecting all FLAC>XXX conversion options except for FLAC>WAV) the correct way to test for this?

y.

That should do it.

Yannzola
2005-07-11, 22:01
Okay...
So I listened to the same track over and over and over again.... with FLAC>>WAV vs. FLAC>>(onboard). Tried it blind (had my wife induge me by engaging/disengaing the toggle and rstarting the track.

Result: I couldn't honestly hear any diffrence. But... I'm not certain that the track was being re-sent to the SB each time. Is there a way to tell what is being sent (WAV or FLAC)?

y.

gorstk
2005-07-11, 22:55
Yannzola wrote:
> Okay...
> So I listened to the same track over and over and over again.... with
> FLAC>>WAV vs. FLAC>>(onboard). Tried it blind (had my wife induge me by
> engaging/disengaing the toggle and rstarting the track.
>
> Result: I couldn't honestly hear any diffrence. But... I'm not certain
> that the track was being re-sent to the SB each time. Is there a way to
> tell what is being sent (WAV or FLAC)?

If you are converting to wav on the server side then falc should appear
in the process list whilst playing the track.

Yannzola
2005-07-15, 18:06
Tried it again... and still can't hear a diffrence. Any others care to give it a go?

y.

Mitch Harding
2005-07-15, 18:34
I have to agree with Sean and some of the others -- there's little to
talk about here without a blind comparison being done. Without that,
there is too much room for error.

On 7/11/05, Timbo <Timbo.1s0q8o (AT) no-mx (DOT) forums.slimdevices.com> wrote:
>
> seanadams Wrote:
> > please back up these claims. You are saying that based on a non-blind
> > subjective listen, our FLAC implementation is *broken*. The correctness
> > of our FLAC decoder is empirical and all you need to do is save the bits
> > at the output to test it.
> >
> > Please note that our implementation:
> >
> > 1) is based on the official FLAC sources - it is the exact same code
> > base as the one that's running on your computer.
> > 2) has been tested by us for bit-perfect output, by recording PCM
> > output back into a computer
> > 3) has also been confirmed by us and others to pass through non-PCM
> > bitstreams correctly
> > 4) has also be tested by the author of FLAC, Josh Coalson, for
> > compatibility with the included test suite
> >
> > There are a couple known bugs and feature requests but none concerning
> > accuracy, which is the entire point of using FLAC!
> >
> > http://tinyurl.com/ak3vc
>
> Hi Sean - well first of all am I correct in thinking that by
> deselecting the FLAC -> FLAC (built-in) and leaving FLAC -> WAV (flac)
> ticked that I am getting PC server side decoding anyway? I could be
> hearing anything making a difference if that isn't the reason?
>
> Secondly who said anything about your decoding implementation being
> *broken*, me-thinks you might be a tad overreacting here. I am really
> talking about the FLAC as compared to WAV issue - just thought the tick
> box made a difference that's all - will test again in a moment.
>
> Rest assured no criticism was intended of the most excellent SB2 -
> after all I have raved about it elsewhere and it is undeniably far too
> good for the price point (put one in a silver hi-fi box with half a
> brick to make it heavier and I would probably have paid 4 times the
> price - and still be amazed at the sound!! :-)
>
> Anyway, sorry if my initial long post seemed griping, not intentional I
> assure you - you're the technician and know about these things - I just
> listen to what my ears tell me and thought someone might be able to
> tell me why they seemed to be hearing what they were hearing?
>
> OK - I am doing this as I write so this is fresh - I won't get into
> built-in v server side decoding yet as I need to make sure I am hearing
> something real...so with all conversion tick boxes as they arrived first
> I compare FLAC to WAV.
>
> I play a FLAC track (Kate Melua - Crawling up a Hill from Call of the
> Search) and it sounds great, nicely turned out and generally fine. I
> rewind and play the first few bars again and then immediately flick on
> to the WAV version that sits below in the playlist (I know, I know but
> I tried doing this blind and just kept ending up playing the wrong
> track!). The difference is harder to describe than I seemed to hear
> last night - there's just slightly more, it's slightly more impressive.
> The best way to describe it is after listening all the way through the
> FLAC track and feeling fine, then the WAV track starts to play and it's
> as though your ears prick-up and take notice - I like it, I can't
> rationalise exactly why but to me there just seems 'more'. Ok so that
> shouldn't be if bits are just bits and data is just data - but to my
> ears and in my system FLAC doesn't do for me what WAV does.
>
> Right - now let's flick the switch on those conversions (leaving only
> FLAC -> WAV flac ticked in the FLAC options) and see what happens
> now...
>
> OK this is more difficult, immediately I thought the FLAC track was the
> WAV track, then I checked and it wasn't, then I rewound and listened to
> the first few bars again and flicked to the WAV track. Now I find it
> more difficult to tell the two apart. So I sit back and listen without
> trying to judge...and my foot is tapping to both tracks, maybe a slight
> edge to the WAV track but this is so slight now as to be easily
> dismissed as placebo...sorry.
>
> OK that's me done...will leave it to the rest of you guys to sort out
> why I am hearing what I think I am hearing - except to say maybe the
> SB2 decoding means the circuit is working harder and so the audio is a
> little less relaxed - I know it's just digits..but..?
>
> All I need to do now is find out how to tag WAV's from EAC so I can see
> them on my SB2? Sorry to cause a stir :-)
>
>
> --
> Timbo
>

m1abrams
2005-07-19, 09:06
What I would like to know from someone however is - bearing in mind I would really like to store my sound files uncompressed - then how do I get EAC to do this and not end up with a WAV file I can’t tag :-/

Why? FLAC produces bit for bit identical to the WAV file, you can test this by ripping a CD to PCM wav, then taking that wav encoding it to flac but save the original wave. Now decode the flac back to wav and perform an md5sum of the two wave files, unless your encoder/decoder is broken they will be identical. This is the procedure that is even on the FLAC website on how to verfiy the encoder/decoder.

With FLAC you get the quality EXACLY the same as WAV plus you get about 2x the storage and 1/2 the needed bandwidth, and you get TAG support.

julian2002
2005-07-19, 09:19
tried it on a few songs (not blind though) and couldn;t hear any difference between internal and external. actually my server is so poor there was a stutter on external (server side) decoding so i'm sticking with internal.
cheers


julian.

Timbo
2005-07-19, 15:34
Why? FLAC produces bit for bit identical to the WAV file, you can test this by ripping a CD to PCM wav, then taking that wav encoding it to flac but save the original wave. Now decode the flac back to wav and perform an md5sum of the two wave files, unless your encoder/decoder is broken they will be identical. This is the procedure that is even on the FLAC website on how to verfiy the encoder/decoder.

With FLAC you get the quality EXACLY the same as WAV plus you get about 2x the storage and 1/2 the needed bandwidth, and you get TAG support.

Hi there - I hear what you say and agree that if space is an issue (or bandwidth if you are wireless) then FLAC certainly adds up to a good idea - but - in my case, in my system, with my ears, I don't *hear* EXACTLY the same quality (and I am not alone) - when it comes to audio quality a few extra quid spent on more HD space (under £70 for a 250gb HD!) is not an issue here.

What goes in may be what comes out when you compare 'bits' - but I have played with this some more over the last few days (I don’t take lightly the format I will choose to archive some hundreds of CDs) and on a well recorded CD track I can spot a FLAC decode instantly on the first few bars. This means for me it’s a no-brainer.

Just let me say obviously if your system isn't able to deliver the subtle nuances involved you will of course not hear any difference - and the system is only as good as it's weakest link - which might be our ears of course (I know mine aren’t what they were after 35 years of interest in the hobby!).

I am not sure what is going on here, but perhaps the analogue audio circuits in my pre-amp may be ultra sensitive to digital noise generated by the SB2 doing it's internal decoding of FLAC sat just above it (it would be interesting to measure this against the same noise generated by internal WAV decoding?). I know that Meridian all digital systems support the digital domain right up until the last moment, only converting to the delicate analogue domain (with all it’s susceptibilities to outside interference) at the very last minute. I prefer the analogue sound however, so my Meridian system contains mostly analogue components.

Who knows the exact reason why I can hear the difference between WAV and FLAC, I certainly don’t, but it doesn’t worry me or cause me anxiety. I am not one of those technical guys who can't envisage anything effecting sound other than that which can be measured - we are talking Hi-Fi here remember. Technical measurements are what design the goods and get us to the gate - but to go through the gate every hi-fi manufacturer understands you have to listen and then fine tune the product with your ears.

If technical measurements were all that mattered in hi-fi then we wouldn’t bother upgrading cables with exotic mixes (and even stranger - exotic cable insulating materials!). We wouldn’t be discussing whether a CD sounds better if you don’t pause it during replay (ok that’s really weird but just ask Jimmy Hughes!), there wouldn’t be such things as turntables with cartridge and arm combinations costing more than my car - and there wouldn’t be those who still prefer glowing valves and a nice set of horn speakers! And I wouldn’t be able to hear a better sounding system after running the Denson De Magic CD - but I can :-)

But please let’s not forget we are talking about a £150 box with an Ethernet adapter sat on top of my hi-fi, streaming digital data from my PC - and it wasn’t long ago I would turn off my PC to listen to my hi-fi as the digital noise down the mains had a negative effect on the sound (this is common practice and if you have one of Russ Andrews mains sniffers you will actually ‘hear’ the noise yourself).

No - this Squeezebox (V2) is unbelievably good and in every respects approaching audiophile quality (and not by accident either if I read the specs right!) so a little debate about file encoding preferences is to be expected in the audiophile forum. Oops, and I have just noticed how long this ‘little debate’ post is - please forgive my rants and I hope at least someone might find my remarks of interest :-)

Mitch Harding
2005-07-19, 15:44
As of your last post, I don't recall a true blind test having been
conducted. Have you had the chance to do that yet?

On 7/19/05, Timbo <Timbo.1sfugb (AT) no-mx (DOT) forums.slimdevices.com> wrote:
>
> m1abrams Wrote:
> > Why? FLAC produces bit for bit identical to the WAV file, you can test
> > this by ripping a CD to PCM wav, then taking that wav encoding it to
> > flac but save the original wave. Now decode the flac back to wav and
> > perform an md5sum of the two wave files, unless your encoder/decoder is
> > broken they will be identical. This is the procedure that is even on
> > the FLAC website on how to verfiy the encoder/decoder.
> >
> > With FLAC you get the quality EXACLY the same as WAV plus you get
> > about 2x the storage and 1/2 the needed bandwidth, and you get TAG
> > support.
>
> Hi there - I hear what you say and agree that if space is an issue (or
> bandwidth if you are wireless) then FLAC certainly adds up to a good
> idea - but - in my case, in my system, with my ears, I don't *hear*
> EXACTLY the same quality (and I am not alone) - when it comes to audio
> quality a few extra quid spent on more HD space (under £70 for a 250gb
> HD!) is not an issue here.
>
> What goes in may be what comes out when you compare 'bits' - but I have
> played with this some more over the last few days (I don't take lightly
> the format I will choose to archive some hundreds of CDs) and on a well
> recorded CD track I can spot a FLAC decode instantly on the first few
> bars. This means for me it's a no-brainer.
>
> Just let me say obviously if your system isn't able to deliver the
> subtle nuances involved you will of course not hear any difference -
> and the system is only as good as it's weakest link - which might be
> our ears of course (I know mine aren't what they were after 35 years of
> interest in the hobby!).
>
> I am not sure what is going on here, but perhaps the analogue audio
> circuits in my pre-amp may be ultra sensitive to digital noise
> generated by the SB2 doing it's internal decoding of FLAC sat just
> above it (it would be interesting to measure this against the same
> noise generated by internal WAV decoding?). I know that Meridian all
> digital systems support the digital domain right up until the last
> moment, only converting to the delicate analogue domain (with all it's
> susceptibilities to outside interference) at the very last minute. I
> prefer the analogue sound however, so my Meridian system contains
> mostly analogue components.
>
> Who knows the exact reason why I can hear the difference between WAV
> and FLAC, I certainly don't, but it doesn't worry me or cause me
> anxiety. I am not one of those technical guys who can't envisage
> anything effecting sound other than that which can be measured - we are
> talking Hi-Fi here remember. Technical measurements are what design the
> goods and get us to the gate - but to go through the gate every hi-fi
> manufacturer understands you have to listen and then fine tune the
> product with your ears.
>
> If technical measurements were all that mattered in hi-fi then we
> wouldn't bother upgrading cables with exotic mixes (and even stranger -
> exotic cable insulating materials!). We wouldn't be discussing whether a
> CD sounds better if you don't pause it during replay (ok that's really
> weird but just ask Jimmy Hughes!), there wouldn't be such things as
> turntables with cartridge and arm combinations costing more than my car
> - and there wouldn't be those who still prefer glowing valves and a nice
> set of horn speakers! And I wouldn't be able to hear a better sounding
> system after running the Denson De Magic CD - but I can :-)
>
> But please let's not forget we are talking about a £150 box with an
> Ethernet adapter sat on top of my hi-fi, streaming digital data from
> my PC - and it wasn't long ago I would turn off my PC to listen to my
> hi-fi as the digital noise down the mains had a negative effect on the
> sound (this is common practice and if you have one of Russ Andrews
> mains sniffers you will actually 'hear' the noise yourself).
>
> No - this Squeezebox (V2) is unbelievably good and in every respects
> approaching audiophile quality (and not by accident either if I read
> the specs right!) so a little debate about file encoding preferences
> is to be expected in the audiophile forum. Oops, and I have just
> noticed how long this 'little debate' post is - please forgive my rants
> and I hope at least someone might find my remarks of interest :-)
>
>
> --
> Timbo
>

radish
2005-07-19, 15:51
So store it as FLAC and stream it as PCM (server side decoding). Then you're getting exactly the same data going to the SB2, thus negating any differences caused by RF coming out of the processor (!?!), but with the addition of easy tagging and 30-40% more available storage. Bargain!


Just let me say obviously if your system isn't able to deliver the subtle nuances involved you will of course not hear any difference

And I'd like to say that I found this phrase rather obnoxious - though I'm sure it was unintentional. The reason I can't hear any difference between FLAC->WAV and WAV is that there IS no difference, it's nothing to do with how much my speakers cost. Mathematics trumps placebo every time.

m1abrams
2005-07-19, 16:06
Hi there - I hear what you say and agree that if space is an issue (or bandwidth if you are wireless) then FLAC certainly adds up to a good idea - but - in my case, in my system, with my ears, I don't *hear* EXACTLY the same quality (and I am not alone) - when it comes to audio quality a few extra quid spent on more HD space (under £70 for a 250gb HD!) is not an issue here.

What goes in may be what comes out when you compare 'bits' - but I have played with this some more over the last few days (I don’t take lightly the format I will choose to archive some hundreds of CDs) and on a well recorded CD track I can spot a FLAC decode instantly on the first few bars. This means for me it’s a no-brainer.

Just let me say obviously if your system isn't able to deliver the subtle nuances involved you will of course not hear any difference - and the system is only as good as it's weakest link - which might be our ears of course (I know mine aren’t what they were after 35 years of interest in the hobby!).

I am not sure what is going on here, but perhaps the analogue audio circuits in my pre-amp may be ultra sensitive to digital noise generated by the SB2 doing it's internal decoding of FLAC sat just above it (it would be interesting to measure this against the same noise generated by internal WAV decoding?). I know that Meridian all digital systems support the digital domain right up until the last moment, only converting to the delicate analogue domain (with all it’s susceptibilities to outside interference) at the very last minute. I prefer the analogue sound however, so my Meridian system contains mostly analogue components.

Who knows the exact reason why I can hear the difference between WAV and FLAC, I certainly don’t, but it doesn’t worry me or cause me anxiety. I am not one of those technical guys who can't envisage anything effecting sound other than that which can be measured - we are talking Hi-Fi here remember. Technical measurements are what design the goods and get us to the gate - but to go through the gate every hi-fi manufacturer understands you have to listen and then fine tune the product with your ears.

If technical measurements were all that mattered in hi-fi then we wouldn’t bother upgrading cables with exotic mixes (and even stranger - exotic cable insulating materials!). We wouldn’t be discussing whether a CD sounds better if you don’t pause it during replay (ok that’s really weird but just ask Jimmy Hughes!), there wouldn’t be such things as turntables with cartridge and arm combinations costing more than my car - and there wouldn’t be those who still prefer glowing valves and a nice set of horn speakers! And I wouldn’t be able to hear a better sounding system after running the Denson De Magic CD - but I can :-)

But please let’s not forget we are talking about a £150 box with an Ethernet adapter sat on top of my hi-fi, streaming digital data from my PC - and it wasn’t long ago I would turn off my PC to listen to my hi-fi as the digital noise down the mains had a negative effect on the sound (this is common practice and if you have one of Russ Andrews mains sniffers you will actually ‘hear’ the noise yourself).

No - this Squeezebox (V2) is unbelievably good and in every respects approaching audiophile quality (and not by accident either if I read the specs right!) so a little debate about file encoding preferences is to be expected in the audiophile forum. Oops, and I have just noticed how long this ‘little debate’ post is - please forgive my rants and I hope at least someone might find my remarks of interest :-)

I am sorry but you have to be kidding me right, is this a troll? Cause it is a good one if it is.

Bits are Bits, either they make it or they dont. You would have to have some serious amount of noise coming from that expensive equipment of yours to cause the bits to be scambled, yet your equipment can still sync on it AMAZING. I am not defending SB right now, just plain old common sense. FLAC is a lossless compresion format and a very good one at that. The author and community take great pride in making sure the codecs for it are up to snuff. Lossless is just that NO LOSS, what goes in comes out the same. It is just plain silly and wasteful to use WAV when a solution like FLAC is available regardless of how much money you can spend on gear.

And my favorite you mention exotic cable, LMAO. Sorry its the engineer in me. Not that this means anything on the Big internet, but I have a degree in Electrical Engineering, currently a software developer, and build my own speakers, and speakers for friends. I build my own speakers not because I am cheap, but because I can.

Do a true blind test and post the results.

Jacob Potter
2005-07-19, 16:19
On 7/19/05, m1abrams <m1abrams.1sfw2n (AT) no-mx (DOT) forums.slimdevices.com> wrote:
> I am sorry but you have to be kidding me right, is this a troll? Cause
> it is a good one if it is.
>
> Bits are Bits, either they make it or they dont. You would have to
> have some serious amount of noise coming from that expensive equipment
> of yours to cause the bits to be scambled, yet your equipment can still
> sync on it AMAZING. I am not defending SB right now, just plain old
> common sense. FLAC is a lossless compresion format and a very good one
> at that. The author and community take great pride in making sure the
> codecs for it are up to snuff. Lossless is just that NO LOSS, what
> goes in comes out the same. It is just plain silly and wasteful to use
> WAV when a solution like FLAC is available regardless of how much money
> you can spend on gear.

In his defense, I will admit that there is a remote possibility that
there could be an audible difference between client-side FLAC decode
and server-side FLAC decode. Higher CPU usage on the SB2 -> higher
power draw -> higher noise -> jitter? Not saying I think it's the
case, just that it's concievable.

But you're right, it is flat-out IMPOSSIBLE for how the data is stored
on the hard drive to make a difference in the sound.

- Jacob

Mitch Harding
2005-07-19, 16:29
I don't think he is claiming that the bits are getting changed en
route to the SB2.

He is claiming that when the SB2 has to do the decoding itself,
perhaps this results in some interference that is audible.

However, I agree that a blind test is required in order to establish this.

On 7/19/05, m1abrams <m1abrams.1sfw2n (AT) no-mx (DOT) forums.slimdevices.com> wrote:
>
> Timbo Wrote:
> > Hi there - I hear what you say and agree that if space is an issue (or
> > bandwidth if you are wireless) then FLAC certainly adds up to a good
> > idea - but - in my case, in my system, with my ears, I don't *hear*
> > EXACTLY the same quality (and I am not alone) - when it comes to audio
> > quality a few extra quid spent on more HD space (under £70 for a 250gb
> > HD!) is not an issue here.
> >
> > What goes in may be what comes out when you compare 'bits' - but I have
> > played with this some more over the last few days (I don't take lightly
> > the format I will choose to archive some hundreds of CDs) and on a well
> > recorded CD track I can spot a FLAC decode instantly on the first few
> > bars. This means for me it's a no-brainer.
> >
> > Just let me say obviously if your system isn't able to deliver the
> > subtle nuances involved you will of course not hear any difference -
> > and the system is only as good as it's weakest link - which might be
> > our ears of course (I know mine aren't what they were after 35 years of
> > interest in the hobby!).
> >
> > I am not sure what is going on here, but perhaps the analogue audio
> > circuits in my pre-amp may be ultra sensitive to digital noise
> > generated by the SB2 doing it's internal decoding of FLAC sat just
> > above it (it would be interesting to measure this against the same
> > noise generated by internal WAV decoding?). I know that Meridian all
> > digital systems support the digital domain right up until the last
> > moment, only converting to the delicate analogue domain (with all it's
> > susceptibilities to outside interference) at the very last minute. I
> > prefer the analogue sound however, so my Meridian system contains
> > mostly analogue components.
> >
> > Who knows the exact reason why I can hear the difference between WAV
> > and FLAC, I certainly don't, but it doesn't worry me or cause me
> > anxiety. I am not one of those technical guys who can't envisage
> > anything effecting sound other than that which can be measured - we are
> > talking Hi-Fi here remember. Technical measurements are what design the
> > goods and get us to the gate - but to go through the gate every hi-fi
> > manufacturer understands you have to listen and then fine tune the
> > product with your ears.
> >
> > If technical measurements were all that mattered in hi-fi then we
> > wouldn't bother upgrading cables with exotic mixes (and even stranger -
> > exotic cable insulating materials!). We wouldn't be discussing whether a
> > CD sounds better if you don't pause it during replay (ok that's really
> > weird but just ask Jimmy Hughes!), there wouldn't be such things as
> > turntables with cartridge and arm combinations costing more than my car
> > - and there wouldn't be those who still prefer glowing valves and a nice
> > set of horn speakers! And I wouldn't be able to hear a better sounding
> > system after running the Denson De Magic CD - but I can :-)
> >
> > But please let's not forget we are talking about a £150 box with an
> > Ethernet adapter sat on top of my hi-fi, streaming digital data from
> > my PC - and it wasn't long ago I would turn off my PC to listen to my
> > hi-fi as the digital noise down the mains had a negative effect on the
> > sound (this is common practice and if you have one of Russ Andrews
> > mains sniffers you will actually 'hear' the noise yourself).
> >
> > No - this Squeezebox (V2) is unbelievably good and in every respects
> > approaching audiophile quality (and not by accident either if I read
> > the specs right!) so a little debate about file encoding preferences
> > is to be expected in the audiophile forum. Oops, and I have just
> > noticed how long this 'little debate' post is - please forgive my rants
> > and I hope at least someone might find my remarks of interest :-)
>
> I am sorry but you have to be kidding me right, is this a troll? Cause
> it is a good one if it is.
>
> Bits are Bits, either they make it or they dont. You would have to
> have some serious amount of noise coming from that expensive equipment
> of yours to cause the bits to be scambled, yet your equipment can still
> sync on it AMAZING. I am not defending SB right now, just plain old
> common sense. FLAC is a lossless compresion format and a very good one
> at that. The author and community take great pride in making sure the
> codecs for it are up to snuff. Lossless is just that NO LOSS, what
> goes in comes out the same. It is just plain silly and wasteful to use
> WAV when a solution like FLAC is available regardless of how much money
> you can spend on gear.
>
> And my favorite you mention exotic cable, LMAO. Sorry its the engineer
> in me. Not that this means anything on the Big internet, but I have a
> degree in Electrical Engineering, currently a software developer, and
> build my own speakers, and speakers for friends. I build my own
> speakers not because I am cheap, but because I can.
>
> Do a true blind test and post the results.
>
>
> --
> m1abrams
>

Timbo
2005-07-21, 05:31
Oh dear well - I think it’s best if I leave this particular topic alone as it seems my audiophile type ‘discussion’ has riled a few folk here and I can really do without that and I know everyone else can as well. I am very happy with my choice of formats (for me) and I was only contributing to what I thought was a interesting thread and offering up my personal findings. So final post...

>there's little to talk about here without a blind comparison being done
>Do a true blind test and post the results
>As of your last post, I don't recall a true blind test having been conducted. Have you had the chance to do that yet?

OK - playlist of comparison tracks - various formats (not just FLAC and WAV) all with same title, sorted into track order (not format). Sit on sofa with SB display turned off, listen to tracks, ‘zapp’ to zapped playlist tracks I feel don’t make the grade (when switching backwards and forwards doing typical comparison as I would do with new piece of equipment or whatever). What’s left was WAV - 100%. This was with built-in decoding. I can ‘barely’ hear any difference (and perhaps there isn’t any difference) when doing this with PC side decoding - but can’t do that blind as need another pair of hands which I don’t have.

>store it as FLAC and stream it as PCM (server side decoding).

That’s actually a very good idea - I did try it and like I said in an earlier post I ‘thought’ I could still tell a difference but wasn’t sure - probably placebo - but space is cheap and honestly, really I don’t mind using the extra space.

> It is just plain silly and wasteful to use WAV when a solution like FLAC is available regardless of how much money you can spend on gear

Please guys don’t try to ridicule me for wasting space - it’s my space and it is cheap and who knows what compression format will be with us in five years time - I am happy uncompressed WAV - you choose FLAC I will choose WAV - no problem.

>Bits are Bits, either they make it or they don’t. You would have to have some serious amount of noise coming from that expensive equipment of yours to cause the bits to be scambled

No, no, no - please, the bits are fine, the noise from the digital source (not the rest of the equipment) would effect the analogue circuitry (in the rest of the equipment). - doh!

> And my favorite you mention exotic cable, LMAO. Sorry its the engineer in me. Not that this means anything on the Big internet, but I have a degree in Electrical Engineering, currently a software developer, and build my own speakers, and speakers for friends. I build my own speakers not because I am cheap, but because I can

What you mean there isn’t a measurable difference when you take two dissimilar metals (or whatever conductor you like) and join them (as in cable to connector via solder to circuit), no impedance change? No reverse electrons zooming back up the wire colliding with those coming down - in that case obviously there is also no advantage from a better or easy conductive load and no need to play with cable structure and build (as in Kimber cables excellent RFI dumping weave) or materials (as in carbon fibre mix or silver for vdh). Good luck with your speakers - try a bit of wet string to connect them to the amp

> The reason I can't hear any difference between FLAC->WAV and WAV is that there IS no difference, it's nothing to do with how much my speakers cost. Mathematics trumps placebo every time.

Rubbish! Math is a tool used to prove a theory - all you have proved is the theory that a WAV file is bit identical to a decoded FLAC file. Now use math to prove that a WAV file decoded on the PC ‘sounds’ exactly the same as one decoded on the SB2. You might have to think out of the box here.

Ok - I'm outa here...:-)

(PS thanks to those who saw I wasn't trying to rock any boats - just stating my findings... :-)

m1abrams
2005-07-21, 06:12
Please guys don’t try to ridicule me for wasting space - it’s my space and it is cheap and who knows what compression format will be with us in five years time - I am happy uncompressed WAV - you choose FLAC I will choose WAV - no problem.

This is my exact reason for using FLAC, because I can easily transcode my entire collection to the next best codec (actually I already transcode to mp3 for use on portables), yes you can do this with WAV but as you have seen tagging WAV is not exactly easy and for my collection having good tags is a requirement which i would assume would be for anyone with a decent collection.

Sorry if I got a little ruffled. You may have found a problem with the SB2 FLAC decoder, however I can not test that because I only have a SQ1. Work with the dev team about resolving the problem, they have been very good at handling such items. However just a point of reference it is pretty much impossible to do a true blind test without assistance, yours is close but not truly blind. Also your tracks used, you had a FLAC and WAV of each track right?

My complaint is your argument that by decoding FLAC to PCM (on any correctly working decoder) there is some how a change in the quality of the data. Which it is just that DATA, and it is been proved over and over that it is the EXACT same data. I do not care what the format original was in, if the data is exactly the same going into the DAC it will produce the same audio out of the DAC. Any difference you can hear is placebo.


What you mean there isn’t a measurable difference when you take two dissimilar metals (or whatever conductor you like) and join them (as in cable to connector via solder to circuit), no impedance change? No reverse electrons zooming back up the wire colliding with those coming down - in that case obviously there is also no advantage from a better or easy conductive load and no need to play with cable structure and build (as in Kimber cables excellent RFI dumping weave) or materials (as in carbon fibre mix or silver for vdh). Good luck with your speakers - try a bit of wet string to connect them to the amp
A marketing persons dream here. Yes measurements are ALL that is needed, their is no voodoo magic in audio or electronics.

My speakers actually are quite good, hooked them up with nice 14ga 2 conductor power cord, used bananna plugs just to make connecting easier since I swap speakers out alot. I have pics of the first set I built, which are based on a design by another fellow. These do not go very low ~80 Hz (small drivers), however the are very precise in the vocal range and highs. Which is good since I made these for my Home Theatre and use a subwoofer to cver the areas below what these speakers can drive. I had to use small drivers because I am married and the wife has final word what can go in the living room, and she does not value function of form as much as I do. My next project will be a pair of nice 2-way bookshelf speakers for the "listening" room.

http://www.thetank.org/gallery/view_album.php?set_albumName=speakerproj

styx
2005-07-21, 06:32
Please read his reply again!


My complaint is your argument that by decoding FLAC to PCM (on any correctly working decoder) there is some how a change in the quality of the data. Which it is just that DATA, and it is been proved over and over that it is the EXACT same data. I do not care what the format original was in, if the data is exactly the same going into the DAC it will produce the same audio out of the DAC. Any difference you can hear is placebo.


No, no, no - please, the bits are fine, the noise from the digital source (not the rest of the equipment) would effect the analogue circuitry (in the rest of the equipment). - doh!
The only plausible reason I can imagine for any audible difference is that increased stress to the PSU affects the voltage or noise to the clocks, which in turn produces jitter. Then again, I'm no electrical engineer, and might be talking out of my ass.

Patrick Dixon
2005-07-21, 06:52
Yes measurements are ALL that is needed,Indeed. All you need to do is to figure out exactly what measurements are required, and then what instruments and techniques you'll need to make them with. As it happens, when it comes to equipment designed to reproduce music, I always reckon your ears are are pretty good substitute.


However just a point of reference it is pretty much impossible to do a true blind test without assistance, yours is close but not truly blind.It sounded pretty good to me. Why do you think it wasn't blind then?

m1abrams
2005-07-21, 08:14
Indeed. All you need to do is to figure out exactly what measurements are required, and then what instruments and techniques you'll need to make them with. As it happens, when it comes to equipment designed to reproduce music, I always reckon your ears are are pretty good substitute.

It sounded pretty good to me. Why do you think it wasn't blind then?

My only real concern is he did not say wether he used the same wav rip for both the wav sample and the flac sample. Also when he loads up the playlist, how does he not know which track is which? If he only has two tracks I guess he could select shuffle, but you always will know which track you started with, and with just 2 tracks not hard to figure which is which. Not saying he did not truly know. But it is generally consider not a true ABX test if done solo. Also a sampling of people is really needed.

pfarrell
2005-07-21, 09:50
On Thu, 2005-07-21 at 08:14 -0700, m1abrams wrote:
> But it is generally consider not a true ABX test if done solo.

Good science usually requires a double blind test, where both the
person doing the test and person administering the test do
not know what is real and what is a placebo.

It would be possible, with enough effort, to make a machine
that could make the randomizing switches, record it, but
it wouldn't be trivial.



--
Pat
http://www.pfarrell.com/music/slimserver/slimsoftware.html

Patrick Dixon
2005-07-21, 10:16
Also when he loads up the playlist, how does he not know which track is which? If he only has two tracks I guess he could select shuffle, but you always will know which track you started with, and with just 2 tracks not hard to figure which is which.I understood that there were a number of tracks in the playlist, each in WAV/FLAC, and then randomly shuffled. Seems 'blind' to me.

m1abrams
2005-07-21, 10:31
I understood that there were a number of tracks in the playlist, each in WAV/FLAC, and then randomly shuffled. Seems 'blind' to me.

But then how would he quickly switch between the 2 same songs that have one as flac and one as wav quickly without looking at the display? Unless he had more than just 1 of each of the same track, that might work. Have like 5 copies of the wav song and 5 copies of the flac song, shuffle them, then play them through and vote on each one. The go back and compare results, however it may be tricky keeping track of which song you voted on, this is why having a second person easier cause then you can label the songs 1,2,3,4,5,etc and then give the tester a sheet labeled accordingly.
This can not be done by the tester because he/she will know that 1 is FLAC, etc.

Yannzola
2005-07-21, 13:53
The only plausible reason I can imagine for any audible difference is that increased stress to the PSU affects the voltage or noise to the clocks, which in turn produces jitter. Then again, I'm no electrical engineer, and might be talking out of my ass.

To me this sounds like a good challenge to all the 'lectrical engineers out there. Are there any =measurable= differences in PSU voltage, clock noise, etc, when the SB decodes a FLAC onboard vs. PCM streamed from the server?

Get out your oscilloscopes, boys.

y.

(definitely talking out of my ass... which, BTW is =not= a good way to break the ice at parties)

Triode
2005-07-21, 14:04
The theory being that the microprocessor executes sufficiently different instructions between decompressing flac and wav to impact the psu in a way that impacts the rest of the player? I believe the processor uses a 1.6V rail and the oscillator impacting jitter is 3.3V so there is little chance of this [even if we believed the first assersion]

Now if this thread had been about the visualizer or scrolling text impacting the sound quality then it would be more interesting.... [withdrawing quietly to see if this sparks some more comparison threads...]

Aylwin
2005-07-22, 01:21
Now if this thread had been about the visualizer or scrolling text impacting the sound quality then it would be more interesting.... [withdrawing quietly to see if this sparks some more comparison threads...]Well now that you mention it, I've noticed between the analog and digital VU meter display that... ;)

Actually, this has been a very interesting thread. When I started out on this journey to audio madness, I was very much into measurements and technical explanations. Over time though I've become less and less so inclined. In the end, what matters is the sound.

I can surely understand Timbo's view point. You try something, you hear a difference, then you try to find out why. Some things you just can't measure or explain but it doesn't mean nothing's actually happening. Personally, I can see no reason why there should be any difference between FLAC vs. WAV or SB2-side vs. server-side decoding. But until I test it myself (maybe one of these days), I can't say for sure.

It should be easy enough to do a blind test:
1. Pick a good quality track and convert to WAV.
2. Using that WAV, convert to FLAC.
3. As suggested by m1abrams, create 5 copies of each.
4. Put them in the same folder.
5. Set the SB2 to play tracks randomly.
6. Browse folder and press play.
7. When you've heard enough, skip to the next track.
IMPORTANT:
8. Only note if the current track is different from the previous.
9. At the end, you should have 9 answers.
10. Go to the PC and check your answers.

Did I miss anything? By the way, this test is based on an article posted by Pat in another thread: http://www.stereophile.com/features/113/

Aylwin

EDIT:
I just noticed (again) the title of this thread: FLAC onboard decoding v. server side in SB2. A similar blind test can be done but requires an assistant. This test will actually be closer to the one described in the article linked above.

DrNic
2005-07-24, 13:35
Okay - a lurker posting here. But not one who isn't interested in attaining the highest quality sound for the most reasonable (wife friendly) cost!!!

Coming from a scientific background (father electronic engineer) myself an orthopaedic surgeon I think you may gather the slant I will have on this thread. But to throw in the curve ball, have any of the audiophiles gone to an audiometrist to have their hearing checked? I mean, alot of the threads I read (here and elsewhere) that insight controversy between "techies" and audiophiles generally end with the statement that "I CAN hear the difference but just can't prove it exists..." This is the whole beauty of music, as it sounds different to us ALL (regardless of system being played back on). Yeah the notes are the same etc but individual perception will never be.
Spending more money on the hardware certainly "levels the playing field" but after a threshold the perceived differences seem to come down to blind faith or the latest fad amongst of the scene.

I would love to see the spectral map of hearing for the audiophile set (and for the techs too) and then lets comment on how good the real "final stage" is in their systems... ! :)

As for FLAC - its the dogs,
server side vs onboard SB2 decoding - still needing the consent to swap out the old SB1!!

Nic

PS: m1abrams - liking your home built speakers BTW!

lostboy
2005-07-25, 02:53
Just in case DrNic's language has confused the non British English users of this forum. The phrase "As for FLAC - its the dogs" is (I think) meant to be short for " ... it's the dog's bollocks" - http://english2american.com/dictionary/b.html refers. Of course I maybe wrong here :-), but I agree with the sentiment as interpreted.

Chris

Mike Hanson
2005-07-25, 03:37
I would love to see the spectral map of hearing for the audiophile set (and for the techs too) and then lets comment on how good the real "final stage" is in their systems... ! :)
Unfortunately, it's not just frequency response that counts here. There's also the sensitivy to the temporal (i.e. timing) elements of the music. Also, a system with a flat frequency response that can't handle dynamics will still sound awful.

Some people are more senstive to the various aspects of music playback than others. Call them "goldears" or whatever. Also, different people will value different aspects in the musical playback (due to their personal sensitivities and expectations, no doubt). As they say, "There no accounting for taste!" <g>

So I don't believe there's a perfect system that would satisfy everyone. We just have to try things out, and decide for ourselves. Ranting back and forth to each other on this board may be entertaining, but beyond that it's pretty useless.

-=> Mike Hanson <=-

DrNic
2005-07-25, 03:59
Just in case DrNic's language has confused the non British English users of this forum. The phrase "As for FLAC - its the dogs" is (I think) meant to be short for " ... it's the dog's bollocks" - http://english2american.com/dictionary/b.html refers. Of course I maybe wrong here :-), but I agree with the sentiment as interpreted.

Chris

Yep - the sentiment is correct!!

And to Mike Hanson - I was not intending to start a "rant" either back or forth!! Just (like everyone else) putting in another view point/opinion... :)
Though human frequency response is fundamental to interpreting music, as a partially deaf person would be hard pushed to detect any difference if they only had a range of 5kHz to 15kHz!! But I agree that the end point is so personal and there is no perfect system for reproducing music (unless we all get a bit acoustic and have the musicians in our rooms - of course setup just as the artist wanted.... -joke :) )

Nic

Mike Hanson
2005-07-25, 05:54
I was not intending to start a "rant" either back or forth!! Just (like everyone else) putting in another view point/opinion... :)
Don't worry, I wasn't accusing you of being heavy-handed or over emotional. <g> I was merely trying to add some perspective to the situation. We may as well be having a conversation about who has the shiniest apple. Of course, it depends on the ambient light, the viewer's sensitivity to the color red, what they ate for breakfast that morning, etc., etc., etc.

We love waxing euphoric about the relative performance of this and that, and I believe we often lose sight of the fact that it's all moot. Sharing ideas is a good thing, but trying to do "online comparisons" is a virtual waste of time. (Or should that be "a waste of virtual time"?)

-=> Mike Hanson <=-

m1abrams
2005-07-25, 06:01
Don't worry, I wasn't accusing you of being heavy-handed or over emotional. <g> I was merely trying to add some perspective to the situation. We may as well be having a conversation about who has the shiniest apple. Of course, it depends on the ambient light, the viewer's sensitivity to the color red, what they ate for breakfast that morning, etc., etc., etc.

We love waxing euphoric about the relative performance of this and that, and I believe we often lose sight of the fact that it's all moot. Sharing ideas is a good thing, but trying to do "online comparisons" is a virtual waste of time. (Or should that be "a waste of virtual time"?)

-=> Mike Hanson <=-

No contest my apple was the shiniest! But the free pony ate it :(

Mike Hanson
2005-07-25, 08:41
No contest my apple was the shiniest! But the free pony ate it :(
My apple will always be shinier, because it's imaginary!

-=> Mike Hanson <=-

Yannzola
2005-07-25, 10:35
Here's a thought: Could it be something like ReplayGain or some such tag info gumming up the works? I hear that server side FLAC's can be "gain adjusted" while onboard decoded FLAC's cannot. Could this be an issue =somehow=.... although I realize that in this instance it doesn't make sense (since the server side WAV and FLAC are identical).

I suggest making certain that the test FLAC's are absolutely tag free during blind testing.

y.

DrNic
2005-07-25, 11:59
Sharing ideas is a good thing, but trying to do "online comparisons" is a virtual waste of time. (Or should that be "a waste of virtual time"?)

-=> Mike Hanson <=-

He he!!
Love it!
Nay probs there Mike.

Nic

Now - letting thread "stay on topic"...

Andrew L. Weekes
2005-07-28, 08:42
The only plausible reason I can imagine for any audible difference is that increased stress to the PSU affects the voltage or noise to the clocks, which in turn produces jitter. Then again, I'm no electrical engineer, and might be talking out of my ass.

You may not be an EE, but you are smart and logical ;)

Andy.

void
2005-08-22, 05:13
The only plausible reason I can imagine for any audible difference is that increased stress to the PSU affects the voltage or noise to the clocks, which in turn produces jitter. Then again, I'm no electrical engineer, and might be talking out of my ass.

You may not be an EE, but you are smart and logical ;)

Andy.

He's on the right track. I'll explain why I think it's almost impossible that FLAC and WAV sound identical with onboard decoding:

Remember that in digital equipment millions of voltage changes occur each second. All parts generate electromagnetic fields, radiofrequencies, causing a lot interference (EMI, RF). Measurments usually show total chaos. Even the most carefully designed equipment still has noise, interferences all over the circuit board(s) and even across equipment. Digital switching puts an aggressive load on power supplies, modulations are unavoidable. Even if all stages are separated and heavily filtered you can still see the digital 'ripples'. These interferences, signal injections and power supply ripples change the switching point (timing) of each bit, which is called jitter. Extremely complex patterns of timing errors (jitter) can be seen over a very wide frequency spectrum.

Now all this noise and jitter isn't a problem at all as long as everything remains 'bitperfect', only for digital audio there's a more serious problem: bits have to be converted to analogue and the audio signal gets distorted when even the smallest amount of jitter reaches the audio conversion. It's not only the clock that gets infected, but the power supply of the DAC and the DAC directly by EMI, RF, remember that it's coupled with the rest of the circuit too (even when using a pulse transformer, etc.).

There's no DAC which can do a jitterless conversion. Even buffering and reclocking isn't a solution, the process itself creates new jitter spectrum that is very similar to the original jitter spectrum, jitter is like a highly infective virus.

I tried many jitter correction devices/solutions including masterclocking from the DAC, etc. but they don't bring the magic, the DAC keeps getting infected.
That's why a very simple DAC can sound very good (my filterless, non-oversampling DAC sounds way better than the very expensive high-end DAC's I used before). That's also why many external superclocks and superregulators are not always the perfect solution, they improve some aspects but very often also introduce new problems. A properly implemented and carefully tuned simple design often sounds (and measures) best.
That's also why good CD-transports are very expensive, it's impossible to get everything right from the optical readout (every correction adds jitter too) to the digital output.

Onboard FLAC conversion has an effect on the jitter pattern/spectrum, but does it sound worse?
The kinds of jitter which vary along with the music signal sound worse than random jitter. I can imagine the FLAC conversion causes a bit pattern which can be correlated to the music signal, so causing a jitter pattern which sounds worse.
(Some designers actually listen to jitter patterns by connecting a small speaker, the correlation easily can be heard because the jitter noise sounds like the music that's played! (heavily distorted).)

Some modifications on the SB2 will decrease jitter (not necessarily clock modifications). However because of the increased resolution it could be that the difference will be even easier to detect. Luckily there's the option of server side decoding.

seanadams
2005-08-22, 16:31
There isn't really a significant difference in total CPU usage for FLAC versus AIFF, or even MP3 for that matter (keep in mind there is a LOT of other stuff going on in the system). The only real difference would be in memory access patterns - here differences is jitter (see other thread - we are talking a change on the order of one PICOsecond) could be explained by power drawn since the IO and memory systems share the rail with the digital audio logic. The CPU core does not - it runs at 1.2V on a separate supply.

John Stimson
2005-08-31, 12:47
I tried many jitter correction devices/solutions including masterclocking from the DAC, etc. but they don't bring the magic, the DAC keeps getting infected.What is the mechanism for upstream jitter to affect the DAC, when the master clock for the system is located at the DAC, on its own power supply?

void
2005-09-01, 02:07
What is the mechanism for upstream jitter to affect the DAC, when the master clock for the system is located at the DAC, on its own power supply?

I'm not an expert, but the jittery data signal has to pass the 'master clock gate' at the DAC. What comes out of the gate is jittery/noisy again.
I guess it's difficult to bring high frequency signals together in one chip, without interacting on each other.

Here's an interesting article from an experienced recording engineer, he doesn't have all the answers (nobody has), but he's onto it:
http://www.digido.com/portal/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=28/

John Stimson
2005-09-03, 14:22
I'm not an expert, but the jittery data signal has to pass the 'master clock gate' at the DAC. What comes out of the gate is jittery/noisy again.
A flip-flop gate has a data input, a clock input, and a data output. Here's how it operates: any time the clock input switches from low to high, the data output is switched to match the level of the data input at that instant. The data output remains at that level until the next low-high clock transition. The time that the data arrives on the input pin does not affect the time that the output pin switches. The same clock signal is used to drive the clock input of every gate in the circuit, so the jitter between gates does not influence the timing of the signal at the final output -- only the jitter on the clock.

Basically, the only thing that affects the signal that you hear is the jitter in the clock at the DAC output. If the clock signal is generated right there, then jitter in earlier stages doesn't cause jitter at the output.

The reason that with SPDIF, jitter in the transport induces jitter in the DAC, if because with SPDIF the clock is generated in the transport.

If jitter at the data input of a circuit influences the timing of the output data, then something is corrupting the clock -- excessive impedance in the power and ground planes, bad signal routing, poor grounding scheme, etc.

void
2005-09-04, 14:46
The time that the data arrives on the input pin does not affect the time that the output pin switches. The same clock signal is used to drive the clock input of every gate in the circuit, so the jitter between gates does not influence the timing of the signal at the final output -- only the jitter on the clock.
In a perfect world... This is how is supposed to work, and it does for most part, but in practice a small amount of jitter/noise leaks through and is difficult to get rid off. The same jitter pattern can be recognised on the output, and it's not coming from the clock.


If jitter at the data input of a circuit influences the timing of the output data, then something is corrupting the clock -- excessive impedance in the power and ground planes, bad signal routing, poor grounding scheme, etc.
Remember that no design or no individual component is 100% perfect (a flip-flop gate is not totally insensitive for noise and voltage fluctuations, etc.). At these high switching frequencies the signals don't look 'digital'.

It's easy to get rid off most jitter, but we heaven't seen or heard any device yet which could get rid off all jitter that you put on its input (I'm not talking about the clock). Several people and manufacturers claim that they can, but each solution we've tried was another dissapointment.
It must be possible though, in theory it seems so easy! It would be nice if someone succeeded because it's this last bit of jitter which seems to corrupt the sound the most.

PhilNYC
2005-09-05, 06:35
Bits are Bits, either they make it or they dont. You would have to have some serious amount of noise coming from that expensive equipment of yours to cause the bits to be scambled, yet your equipment can still sync on it AMAZING. .

Bits are bits, but in the end, the digital to analog conversion process is still an electrical/analog process. Variations in the squarewave voltage that represents a bit, while accurately representing a 1 or 0 within the tolerance of a computer's design, can introduce variances in the analog waveform. Resulting effects include jitter (timing errors) and amplitude errors. Is it audible? Depends on the system, the amount of jitter, etc. I've done double-blind tests on things like transports...both of which delivered identical bits, but had measureable differences in jitter, and resulted in audible (sometimes significant) differences in sound. And this was done using a DAC that had a very sophisticated buffering and re-clocking design...