Emu PatchMix experts inside...

[adhoc]

Quote:

Originally Posted by Solude ...

you've raised something else rather important imo - what exactly does replaygain do to the sound, bitwise?

 

[Solude]

Well that bites, the 0dB is correct and the -6dB put on during CD playback is artificial. That means sadly that the lossless copy is a truer version than the original when played on a PC

 

[thomaspf]

Lossless is of course lossless and nothing else. Whether it comes from a PC or not does not matter.

If there is anything mocking with the bits that is a problem of the software/driver you are using for playback and not a problem for the PC in general. I can assure you that my ripped tracks on the PC generate the same samples as myCD player.

As a couple of posters have reported it seems very difficult to get the bit perfect playback with the EMU cards.

Cheers

Thomas

 

[Erukian]

Ok, so tell me if i messed up something with this

E-MU's run at 24bit, doesn't matter if your giving it 16bit or 32bit material, patchmix will make it 24bit. So in theory, if you use a program like foobar and have the program work everything out in 32bit, then patchmix rounds off some of the 0's from your trim and you'll get 24bit or pads the 16bit material to 24bit (lossless either way).

I can't see a loss of bits (resolution) in the main signal causing a loss of quality. Someone should try a -30dB trim and run it through digital out then back in to check it's bit-perfectness.

-Joe

 

[thomaspf]

If your source material is 16 bit and you extend this to 24 bits or 32 bits then the best non destructive conversion is to put the orignal 16bits of data into the most significant bits of the extended samples.

Any +6db boost will drive this signal into clipping since many CDs are mastered rather ot these days.

If you apply a shift operation by 5 bits (-30db gain) then you should find all the bits in the resulting stream. However when you convert this signal with a DAC then the low bit is now at position 21. This sits in the noise floor of most DACs.

Hope this answers your question.

Cheers

Thomas

 

[Solude]

Not sure Thomas is getting it so I'll rephrase.

The Lossless copy of the CD is a bit perfect copy of the original, i.e. provides the same gain as through a stand alone CDP.

If the CD used to make the Lossless file is played on the PC, it will have its gain reduced by -6dB consistently, everytime.

Odd but that's how Emu or Windows behaves. No trims, volume adjustments, dsps, processes or any other junk just straight play... -6dB if the source is the CD and not the Lossless file created from it.

Another way to put it is if I play a CD in my Sony standalone it puts out 0dB. If I play the Lossless WMA made from it, again its pegged at 0dB. If however I play that CD in the PC it gets dropped to -6dB with no user intervention.

That make more sense? I'm not questioning it, just making others aware since I would not have thought CD playback wasn't bit perfect. In short if you want bit perfect playback use a lossless codec and not the CD itself.

 

[thomaspf]

When you say:

Quote:

If the CD used to make the Lossless file is played on the PC, it will have its gain reduced by -6dB consistently

What do you actually do to play that CD? My CDs play just fine with bit perfect out.

Cheers

Thomas

 

[Solude]

Nothing special. Place CD in drive, open WMP, select CD and it goes.

WMP is everything off, full volume, CD set to 24bit for HDCD.

PatchMix is Minimum -10 set, optical out switched to S/PDIF, metres removed, WAVE send to S/PDIF.

That's it. Result Lossless 0dB, CD -6dB.

 

[thomaspf]

Well, try a player that does nothing but digital extraction and bit perfect playback.

CD playback in WMP is a special case. Have you ever tried a player like Jriver Media Center with ASIO output. If you can live with basic comfort have a look at foobar.


Cheers

Thomas

 

[Cthulhu]

Quote:

Originally Posted by adhoc you've raised something else rather important imo - what exactly does replaygain do to the sound, bitwise?

Nothing. replaygaining a file inserts metadata into the file header to tell the software system how loud to play it back.

If you mean mp3gaining files for things that don't support replaygain properly (I'm looking squarely at ipods here) mp3's are encoded in frames, and as I understand it, each frame has volume infomation attached. MP3gain goes through and changes the volume infomation with each stage.

Reading these tags is handled by your mp3 decoder, which then spits out the unencoded audio stream to the next item down the line.

 

[Solude]

Again not looking to 'fix' this behavior. Just nice to know about it. Playing library Lossless WMA is far more convienent than grabbing for a CD every little while

 

[Jasper994]

Quote:

Originally Posted by Solude Have to ask, Jasper, why such odd -dB settings? Why not just a straight -9dB or -5dB?

When you pull down the patchmix slider it likes to lock on to odd increments... <shrug>

 

[Erukian]

So one last time (for the sake of over-clarifying) if i have foobar set to 32bit, that's 16 bits of padding for all my non-dvd-audio material.

If 5 bits is -30dB, then someone can lower the volume about -100dB before it starts cutting into the original 16bit resolution - or am I wrong?

-Joe

 

[adhoc]

Quote:

Originally Posted by Cthulhu Nothing. replaygaining a file inserts metadata into the file header to tell the software system how loud to play it back.

nope, i use only replaygain with my flac files.

going along with the general spirit of this thread, wont lowering the volume digitally via software, which is essentially what replaygain does for me, lead to loss of bits?

very very very confused here. it seems that applying any form of digital attenuation would affect the bits.

 

[thomaspf]

Erukian,

if you take a 16bit signal and stuff it into the 16 most significant bits of a 32 but sample you can attenuate (or shift) this by 16 bits without loosing any information in the digital domain.

However, your DACs only work with 24bit samples. when your 32bit samples attenuated by 100db are handed to the DAC chip for conversion you now have only the lowest 8 bits in use. The real resolution of your DAC is something around 17-18bits before noise so on playback you now only have 2-3bits of significant data before noise.

Cheers

Thomas