Win 7 Optimisations

[Phil Leigh]

It seems everybody is talking nonsene except you, dear Phil. Your attitude is getting more and more annoying to be honest.


1. The recording usually introduces losses and phaseshifts.
That recording is base to calculate the filters. And these are
multiplied with the data.
That's the first time you introduce losses.
2. You have to redo those everytime you change sometging in the system. Or you got?? Correct!! "Losses".
3. The filter 100% matches your recording position. Half a meter to the left
or even your left and right ear won't have that exact position of your
mike anymore. "Losses!!"
4. No filter is lossless. You introduce ringing and partly phaseshifts
to the signal. "Losses" ( Lets skip 32 vs 64 bit calculations)
5. The filter resolution or number of taps makes a difference. Losses.
6. Different samplerates make a diiference. "losses".
7. To avoid clipping you at least apply 3db attenuation (prior to
convolution) on usually 16 bit base data, which are already dithered.
"Losses"
8. To do all the processing you need quite some processing power.
Especially if you also run a crossover.
That'll cause jitter and noise again. "Losses"

I'll stop here. Because that are the losses I can put together within a minute of thinking about it.
Even on the DRC page you find more of those.

Now you explain where you discovered a lossless convolution. I bet you'll win the Nobel price for that finding.

Phil. Sorry. You're wrong. Face it.


P.S: Above doesn't mean that the "losses" introduced by room resonances are
irrelevant. It's the other way around. Still convolution is and will remain
a compromise solution. To many people the effects of convolution
outweigh by far the losses introduced by doing it. The worse your setup and room
the more you'll like it.
I prefer to run without it.

Sigh...

the processing/convolution is not done in 16-bits
I resample everything to 96k for processing.

Most people I know that use DRC find that the beneficial effect of correctly targeted DRC in typical domestic rooms outweigh the issues you raise. Clearly we know different people with different equipment and different rooms.



I agree I should not have used the term "lossless". Obviously the process changes the bitstream - for the better in this case,in my opinion.

My active crossovers are implemented in the analogue domain.

[Phil Leigh]

and by the way, if I need to do any CRITICAL listening I use my headphones. They are of superior quality to any speaker system available and I don't need DRC...

So stop trying to sidetrack this discussion onto the merits of my replay chain.

[soundcheck]

Sigh...

the processing/convolution is not done in 16-bits
I resample everything to 96k for processing.

I overead your "sigh". Stands for your attitude.

You got a logical problem in above. You mixed up bitdepth and samplerate. That can happen.

And guess what: Resampling introduces - - Guess what??

Yep. "Losses"

And even if you generate 24bit files. The nasty dither of 16bit
material stays in the signal and gets convolved.

[soundcheck]

and by the way, if I need to do any CRITICAL listening I use my headphones. They are of superior quality to any speaker system available and I don't need DRC...

So stop trying to sidetrack this discussion onto the merits of my replay chain.

You gotta seriuos problem with your attitude.

Calling other peoples findings nonsense. Without being able to prove what your're saying and even coming up with false claims.

And finally trying to cut the discussion by a more than arrogant "Sigh".

No Phil. It's not working like that.

And I'm sure the next time you'll post again your believes. Because I think we've had a similar disussion about convolution before.


Anyhow this discussion is another example that underlines what somebody called earlier "bold" statement, that I made earlier in this thread.

I'm outta here. Not worth it.

[Phil Leigh]

I overead your "sigh". Stands for your attitude.

You got a logical problem in above. You mixed up bitdepth and samplerate. That can happen.

And guess what: Resampling introduces - - Guess what??

Yep. "Losses"

And even if you generate 24bit files. The nasty dither of 16bit
material stays in the signal and gets convolved.

I'm not confused about sample rate and bit depth thank you. They were 2 seperate points.

And I'm not geenrating any files - the resampling(to 24/96) is done on-the fly via SOX. I convert everything to 24/96 (24-bit source files do not have their bit-depth altered by this process) before it is piped to Inguz so that Inguz can use a 24/96 optimised convolution target file at the same frequency/bitdepth. This actually reduces the Inguz CPU overhead, as Inguz never needs to internally resample the convolution file to match the music.

As for the nasty 16-bit dither... well, any lossless chain is going to present that dithered half-bit to you just as it is, right?

This is all off-topic.

Discuss why I can't hear the impact of server priority/performance changes via my headphones... without suggesting my hearing is faulty, I had it tested 2 months ago and it is normal.

[darrell]

http://www.empiricalaudio.com/computer-audio/

That advice (be it good, bad or indifferent) is related to the computer as a player, outputting via USB to a DAC. This is not the architecture of a Squeezebox solution. The computer running Logitech Media Server is not playing music - it is simply transferring data to the Squeezebox appliance. It is the Squeezebox which is the player.

In another computer music solution (not using LMS or a Squeezebox), one might have ones music files stored on a NAS, and play them back using a software music player (Foobar or whatever) on a laptop connected to a hifi system. In this architecture, the NAS is equivalent to our LMS server, and the laptop is equivalent to the Squeezebox, so tweaking the laptop's settings/priorities/whatever might make some sense.

There still seems to be a lot of confusion about this.

[chill]

There still seems to be a lot of confusion about this.

Darrell - yes, that was worth pointing out again I think. I suspect there are only one or two people who are confused about this at this stage, but I can't help feeling that new/casual visitors to this forum will see this thread, and its apparently legitimate sources, and conclude that they need to agonise over setting up their servers. They might also be put off all the excellent server alternatives (NAS, plug computers, mac mini etc) because they won't run Windows 7 and can't be set up with these audiofool tweaks.

[Phil Leigh]

You gotta seriuos problem with your attitude.

Calling other peoples findings nonsense. Without being able to prove what your're saying and even coming up with false claims.

And finally trying to cut the discussion by a more than arrogant "Sigh".

No Phil. It's not working like that.

And I'm sure the next time you'll post again your believes. Because I think we've had a similar disussion about convolution before.


Anyhow this discussion is another example that underlines what somebody called earlier "bold" statement, that I made earlier in this thread.

I'm outta here. Not worth it.

you are the one with an attitude problem.
You leapt onto this thread.
I 'm not the one making claims, false or otherwise.
as usual you have run away without answering my question. All you ever do is jump on me whenever I post anything, make some gratuitous remarks and then run and hide.

So far there has not been a single shred of evidence to support the preposterous claim that the bits in the buffer can be AT THE SAME TIME be both correct values and somehow the "wrong shape".

That is because it is utter nonsense and cannot be supported by any facts.

I get all the flac (sic) around here because I'm one of only a few people seemingly brave enough to call you out on this. not surprising given the insults that come flying back if you dare to do this.

I am a perfectly reasonable person who is open to ideas. I've tried all the things suggested and they don't make any difference to me. I've studied the so-called science of them and they don't stand up to any level of rational scrutiny.

The moment I start to get close to forcing the argument to a resolution, we get either:
1 - you must be deaf because all My golden-eared friends can hear the improvements or
2 - your system/setup is flawed and you will never hear the improvements

It's time people seriously considered that there might be an option 3...

[garym]

It's time people seriously considered that there might be an option 3...

Agree. And I think option 3 has been pretty clear for quite a while.

[kakklank]

I'm outta here. Not worth it.

Is that a promise. Nothing would please me more, and while you're at it please take SBGK with you. Maybe add a forum to your blog so you can continue your audiophool discussion there.