[Ham Sandwich]

My philosophy about this is to set up a system (hardware and software) that is capable of playing bit-perfect.  Then once that is achieved I fiddle and do things like digital EQ, resampling, digital volume adjustment like replay gain, a slight digital volume decrease to prevent peaks from going over full scale, things like that.  So I go through the works and setup to make things bit-perfect then immediately mess it up.  And that's the way I want it usually most of the time.
 
If you happen to have a DAC that can decode HDCD and you play a ripped file that has HDCD encoding then that is one time where you want (need) bit-perfect all the way through to the DAC.  If the software messes with the bits and isn't bit-perfect then the DAC won't see a valid HDCD signal and you won't get the HDCD decoding.  If the software does everything bit-perfect then the DAC will see the HDCD signal and decode it.

 

[erozsolt]

Can someone explain me why is it good to do a -3 db gain to a lossless source? If I play a Redbook CD in a many thousand dollar expensive TRANSPORT, then what happens exactly? I think there is no -3 db gain at all! Or is there?
 
I don't get that for example if there is any SPDIF signal between the computer and a DAC, then why should we limit the music, or apply replaygain, if for 20 years externals transports didn't need to do that! Can someone explain what is a difference between a computer playing lossless audio over SPDIF and a redbook transport playing a CD and outputting to SPDIF?

 

[RonaldDumsfeld]

 
Quote:

Can someone explain

 
Someone already has to a certain extent. You need to read the links in post #4 or google 'inter sample peaks' .
 
There is nothing wrong with the optical drives, media players, DAC. It's not the equipment. It's the CDs themselves. The Mastering Engineers have compressed them so hard in order to get maximum  volume that some of them actually do clip when the original wave form is recreated. Light clipping isn't as obvious as you might think but it's still advisable to give yourself some headroom if you can. 

 

[erozsolt]

OK, but even if the problem does exist, what you say is that all external CD transports or any standalone CD players must have the same error, don't you? They usually have no circuit before the DAC chip which would do any kind of digital volume or gain control. So what you say is that every high-end transport + DAC combo would clip with modern CDs?
 
I might read into it, but for me, it's a highly unlikely situation.

 

[estreeter]

I dont pretend to understand half the technical issues here, but when Gordon Rankin tells me that USB is a perfectly acceptable medium for transmitting data to a DAC, I tend to take notice. Maybe its the old code-cutter in me, but I respect anyone who has burnt the midnight oil to build a better digital mousetrap.

 

[erozsolt]

I have read the article in post #4. For me, it's clear that any kind of well mastered CD would never produce this clipping. In the part where it says that "how to prevent..." it says that it's a one click switch in Ozone for example, so i would think that this whole problem is part of every undergraduate audio engineer course, and all the mastering softwares have a one click way of preventing it happening. On my side, I will just continue listening to nice, raw, untampered, unmodified, non replaygained sound from the computer, just as if it was a CD transport.
Quote:

 
 
Someone already has to a certain extent. You need to read the links in post #4 or google 'inter sample peaks' .
 
There is nothing wrong with the optical drives, media players, DAC. It's not the equipment. It's the CDs themselves. The Mastering Engineers have compressed them so hard in order to get maximum  volume that some of them actually do clip when the original wave form is recreated. Light clipping isn't as obvious as you might think but it's still advisable to give yourself some headroom if you can. 

 

[anomalouscode]

think of text instead of audio.
 
back in the day, songs were engineered to lets say about 12pt.
 
then one day some guys came out with a song at 23pt, which made the 12pt's sound smaller and less powerful.
 
so everyone started to up the size of the font to stupid numbers, to the point where the song would barely fit on the page, but when listened made their font more powerful than the next mans.
 
now, raising the gain digitally is the same as increasing the size of the font in word, but having to keep the whole thing on the page no matter how big the font... where as turning a volume knob is like using a magnifying glass instead of increasing the font.
 
increasing the font changes the font and the document itself, where as a magnifying glass doent change the font, but rather how you see(hear) the font.
 
so if you want to hear the record as intended, whether well mastered or otherwise, use your volume knob, thats what its there for.
 
my crappy analogy  2c 

 

[erozsolt]

Of course we use the volume knob, but all what you say is just the phenomenon that classical music albums are usually much more quiet than pop CDs. But what I say is that if you can only have 16-bits to play with, then you shouldn't alter the sound digitally, you should only put any volume adjustment after the DAC. On the other hand, with a 24 bit DAC, you can even use just 25% of the available volume range, you will still have 22 bits to play with, which is enought for anything.
 
So I would say that listening @ bit-perfect is only important for 16/44.1 DACs, but for 24/96 you only need to be sure that your system is capable of bit-perfectness, after it, you can just tinker with it. Like RG, DSP, using the volume control in foobar, etc. For 16-bit playback, I would strictly stick to unaltered data.
 
So my rule of thumb:
- for a 16-bit DAC or a USB 1.1 solution only capable of transfering 16-bit music, I would just use KS/ASIO/WASAPI + no RG, no DSP, foobar volume 100%, windows volume 100%
- for a 24-bit DAC or a USB 2.0 solution capable of transfering 24-bit music, I would just use KS/ASIO/WASAPI and it doesn't matter what you do in a player or with the volume controls
 
 
 
Quote:

think of text instead of audio.
 
back in the day, songs were engineered to lets say about 12pt.
 
then one day some guys came out with a song at 23pt, which made the 12pt's sound smaller and less powerful.
 
so everyone started to up the size of the font to stupid numbers, to the point where the song would barely fit on the page, but when listened made their font more powerful than the next mans.
 
now, raising the gain digitally is the same as increasing the size of the font in word, but having to keep the whole thing on the page no matter how big the font... where as turning a volume knob is like using a magnifying glass instead of increasing the font.
 
increasing the font changes the font and the document itself, where as a magnifying glass doent change the font, but rather how you see(hear) the font.
 
so if you want to hear the record as intended, whether well mastered or otherwise, use your volume knob, thats what its there for.
 
my crappy analogy  2c 

 

[FallenAngel]

I'm sorry, this makes no sense as you're going off a wrong assumption - just because content has 24-bit depth doesn't mean it's not just as compressed and the amount of resolution you're left with for volume control can be just as small.
 
Quote:

Of course we use the volume knob, but all what you say is just the phenomenon that classical music albums are usually much more quiet than pop CDs. But what I say is that if you can only have 16-bits to play with, then you shouldn't alter the sound digitally, you should only put any volume adjustment after the DAC. On the other hand, with a 24 bit DAC, you can even use just 25% of the available volume range, you will still have 22 bits to play with, which is enought for anything.
 
So I would say that listening @ bit-perfect is only important for 16/44.1 DACs, but for 24/96 you only need to be sure that your system is capable of bit-perfectness, after it, you can just tinker with it. Like RG, DSP, using the volume control in foobar, etc. For 16-bit playback, I would strictly stick to unaltered data.
 
So my rule of thumb:
- for a 16-bit DAC or a USB 1.1 solution only capable of transfering 16-bit music, I would just use KS/ASIO/WASAPI + no RG, no DSP, foobar volume 100%, windows volume 100%
- for a 24-bit DAC or a USB 2.0 solution capable of transfering 24-bit music, I would just use KS/ASIO/WASAPI and it doesn't matter what you do in a player or with the volume controls
 
 
 
Quote:

 

 

 

[erozsolt]

Mathematically you are right, but human ear is not resolving over 18-19 bit. And even in a high resolution music, the bits over 20 bit are mostly just noise. That's why for example on a 24 bit music, dithering has no effect because you are essentialy modifying the 24th bit, which no one can hear, or no equipment can really reproduce.
 
But my example was for listening Redbook CDs over 16-bit vs. 24-bit. If you listen to a Redbook CD over 16-bit and use a foobar volume control at 50%, then you have already reduced your usable bits to 15-bit! Thats a big difference, you can easily A-B test 15-bit vs. 16-bit.
 
If you do the same on a 24-bit equipment, then as a start you have 16-bit usable information and 8-bit with 0-s. If you apply a digital volume control in a 24-bit system, you only loose 1-2 bit from 24, not from 16, so essentially you will have a system capable of playing back 22-23 bit musical information, which is still way higher than what you can record/reproduce/hear, which is about 18-19 bit.
 
 
 
 
Quote:

Of course we use the volume knob, but all what you say is just the phenomenon that classical music albums are usually much more quiet than pop CDs. But what I say is that if you can only have 16-bits to play with, then you shouldn't alter the sound digitally, you should only put any volume adjustment after the DAC. On the other hand, with a 24 bit DAC, you can even use just 25% of the available volume range, you will still have 22 bits to play with, which is enought for anything.
 
So I would say that listening @ bit-perfect is only important for 16/44.1 DACs, but for 24/96 you only need to be sure that your system is capable of bit-perfectness, after it, you can just tinker with it. Like RG, DSP, using the volume control in foobar, etc. For 16-bit playback, I would strictly stick to unaltered data.
 
So my rule of thumb:
- for a 16-bit DAC or a USB 1.1 solution only capable of transfering 16-bit music, I would just use KS/ASIO/WASAPI + no RG, no DSP, foobar volume 100%, windows volume 100%
- for a 24-bit DAC or a USB 2.0 solution capable of transfering 24-bit music, I would just use KS/ASIO/WASAPI and it doesn't matter what you do in a player or with the volume controls
 
 
 
Quote:

 

 

[anomalouscode]

regardless of 16 or 24, if you want to listen to the music as it was intended, which alot of audiophiles do, then you shouldnt be doing anything at all to anything except turning the volume knob.
 
for classical, if it sounds lower, its lower in this section to contrast with the massive dynamic range of the music which can go from a whisper to cannon-loud in a split second,... that is part of the composition and therefore part of the recording. 
 
now if you gain it all up and compress, you're pulling down the dynamic range and not getting the full effect,... a whisper should not be equal in loudness to a cannon. but it is in pop music where you'll get gunshots next to some chick singing,... which is not natural, but then pop music isnt going for natural.
 
so regardless if you have a few bits to play with or not,... if you alter the file digitally through a player or a DAW, before the DAC, you're messing with the material.
 
whereas if you do it after, provided you dont have a compressor in line like some might, then there's no messing. 
 
but i think it really depends on what you listen to,... classical, yes you want the full, as it was intended feeling,.... but speed black metal, hip hop, top 40 stuff, sure, compress it some more, you probably wont really notice at all.

 

[Vitor Machado]

Quote:

If you do the same on a 24-bit equipment, then as a start you have 16-bit usable information and 8-bit with 0-s. If you apply a digital volume control in a 24-bit system, you only loose 1-2 bit from 24, not from 16, so essentially you will have a system capable of playing back 22-23 bit musical information, which is still way higher than what you can record/reproduce/hear, which is about 18-19 bit.
 

I don't understand this very well... Suppose you have 100 in 16-bit, that is, 00000000.00000100, and then you pad it to 24-bit, that is, 00000000.00000000.00000100. In both cases, if you reduce 100 from these values, you'll reach zero. You do get a higher range with 24-bit, but are you sure we'll be losing less when we reduce the volume of a 16-bit source padded to 24-bit?

 

[erozsolt]

You are almost there, except that reducing volume is something like multiplying with a number between 0 and 1, let's say 0.8. And think about padded bits as decimal places available. So at 16-bit padded to 16-bit you have only integers. But at 16-bits padded to 24-bit you have room for the fraction part. It's just an analogy.

 

[Roseval]

16 bit:
00000000.00000100, 
then you pad it to 24-bit,
00000000.00000100.00000000
 
Apply volume reduction
00000000.00000100.00000000
000000000.00000100.0000000
0000000000.00000100.000000
00000000000.00000100.00000
000000000000.00000100.0000
0000000000000.00000100.000
00000000000000.00000100.00
000000000000000.00000100.0
0000000000000000.00000100.
 
With this (very primitive ) volume control we simply chop 8 bits off to obtain (6*8=) 48 dB volume reduction without losing resolution
 
 
 
 
 

 

 

[Vitor Machado]

 
Quote:

16 bit:
00000000.00000100, 
then you pad it to 24-bit,
00000000.00000100.00000000

 

Oh I should have thought, the bits are added to the end of the sequence instead of the beginning...
Thanks.