Emu PatchMix experts inside...

[Cthulhu]

Quote:

Originally Posted by adhoc nope, i use only replaygain with my flac files.

What this has to do with anything.

The proposed replay gain standard clearly indicates that it works by inserting metadata into files (APEv2 tags) and they are used by devices to determine the correct loudness at which to playback the file.

Why would using FLAC or any of the numerous other formats supported by the replay gain spec make a difference here? now the specific program MP3Gain has some other (funky) options, but that is not the discussion at hand.

Quote:

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?

No, because replaygain does it's magic when decoding FLAC -> PCM

Thats what i was trying to say in the first instance. The other digital attentuation methods here work after the file has been decoded, the replaygain metadata tags merely give additional infomation how to generate the PCM waveform (by changing the position of the waveform relative to 0, so the average volume is 89 DB instead of 98,for example)

 

[adhoc]

Quote:

Originally Posted by Cthulhu No, because replaygain does it's magic when decoding FLAC -> PCM Thats what i was trying to say in the first instance. The other digital attentuation methods here work after the file has been decoded, the replaygain metadata tags merely give additional infomation how to generate the PCM waveform (by changing the position of the waveform relative to 0, so the average volume is 89 DB instead of 98,for example)

that's just what i needed to know - i apologise if my ignorance has ruffled any feathers. thanks for that clarification cthulhu!

 

[Solude]

Quote:

Originally Posted by thomaspf First -25db is a bit more than 4 bits of resolution that you are now pushing down into lower significant bits which translates into lower voltages on the input of your amp. Effectivley you have limited the voltage swing on the input of your amp. To compensate for that you now have applied maximum gain. This will of course amplify not just your dynamically compressed signal but also the noise floor with maximum gain.

I was thinking about this earlier today and it stuck me that if this was a speaker rig with separates this behavior is par for the course. All preamp/amp combos do this. You pass full signal to the pre, attenuate it dropping the voltage level and pass it along to the amp.

In our scenario, the headphone amp is really an integrated and the lower level only exists over a few inches of cable. So we think the lowering the level at the DAC and passing it through an interconnect is harmful to the result. But audiophiles have done this forever in speaker rigs.

So in hind sight I don't think reducing the voltage level to the amp(integrated) by a reasonable ammount is harmful to the result any more than passing a lower level to a separate amp in a preamp/amp speaker rig is.

I'll still be sending my Reference to HeadAmp to have the gain changed to 3x/9dB though

 

[doctorcilantro]

Very interesting thread guys.

I was confused for a second, since I use Replay Gain with my lossless files so when I pop a cd in the HTPC I get a much louder output, simply because Replay Gain is controlling my db levels on the lossless stuff.

Now I find it very strange the WMP would actually attenuate a cd that has already been recorded at maybe -6db to -3db and then mastered up to -.2db.

Sounds like you should try J. River or Foobar and see what happens. Isn't the default that ReplayGain is shooting for something like -89db output? I try to boost that by a fixed amount (most files are reduced by at least 10db) because my Pre-amp is also attenuatiing the signal before sending to my monos. The poster who was trying to get the signal hotter so as to be in a more usable range seemed right to do so. I read some documentation of my preamp and it stated that you would get further away from the noise floor of the pre.

DC

 

[Tim D]

Quote:

Originally Posted by Solude I was thinking about this earlier today and it stuck me that if this was a speaker rig with separates this behavior is par for the course. All preamp/amp combos do this. You pass full signal to the pre, attenuate it dropping the voltage level and pass it along to the amp. In our scenario, the headphone amp is really an integrated and the lower level only exists over a few inches of cable. So we think the lowering the level at the DAC and passing it through an interconnect is harmful to the result. But audiophiles have done this forever in speaker rigs. So in hind sight I don't think reducing the voltage level to the amp(integrated) by a reasonable ammount is harmful to the result any more than passing a lower level to a separate amp in a preamp/amp speaker rig is. I'll still be sending my Reference to HeadAmp to have the gain changed to 3x/9dB though

But the difference is digital attenuation modifies the signal and throws bits (or the low level signal components) away entirely while the preamp just attenuates the signal.

Assuming we have a quality pre-amp that does not add noise or degrade signal (this is not always real world performance, but still usually better than throwing a large amount of bits away), if we attenuate than amplify we should still have most of the original signal (because the low level signals are still there, albeit at even lower levels). If we attenuate digitally and the process does not take advantage of over 16-bit headroom, when you amplify the result you will not get the low level details back. Digital attenuation is fine if you do *not* significantly amplify after the DAC.

All of this can be verified pretty easily including how much headroom your DAC or digital processing has. For example I believe you should be able to create a low level test tone around -70db. When amplified you should be able to easily hear it assuming your system doesn't have poor S/N. Now apply about -26 to -30db digital attenuation and see if any additional amplification retrieves the low-level signal back. And if it does than that means you either have plenty of headroom in DAC or processing, or the trimpots are actually analog attenuators that are digitally controlled, etc. Not to mention if you can just output your digital signal to disk, you should be able to verify what is happening with any sound editor.

Or just think about resizing a digital image to be smaller. When stretched out again you will lose quality compared to the original. However an actual optical lens is an 'analog' way of resizing the signal (likewise the quality of lens matters in final quality).

Also what bit-depth is used for processing is really crucial to wheter it all matters. It isn't difficult to find active analog audio circuits that can resolve beyond 16-bits or 96db S/N, however it *IS* difficult or perhaps even impossible to find any that can truly resolve full 24-bits or 144db.

 

[thomaspf]

Here are some interesting thoughts on a transformer based volume control and how it differs from the more usual stepped resistor network.

http://www.audio-consulting.ch/

I have no experience with this at all, just list it for completeness of the discussion.

Cheers

Thomas

 

[doctorcilantro]

Interesting (see below):

"Using digital attenuation to max out the volume on your amp is very likely not the correct strategy.

First -25db is a bit more than 4 bits of resolution that you are now pushing down into lower significant bits which translates into lower voltages on the input of your amp. Effectivley you have limited the voltage swing on the input of your amp.

To compensate for that you now have applied maximum gain. This will of course amplify not just your dynamically compressed signal but also the noise floor with maximum gain."

My issue is what's going on with the EMU DSP (1212M). I send the signal from a Sunfire Preamp (phono stage is (2) 12AX7's) to the analog inputs of the 1212M and I could use more gain. I can do a resistor change to get some more but I don't have the time or the skill, at least right now. Wavelab is set to record with 32bit float; so am I losing something going through the EMU if I add some Trim Pots to my ASIO sends? I was also using the AUX ouput with +7.7 db to cover the Replay Gain reduction which averages about -12db (since I want ALL tracks to be at the same volume not just albums). But that's beside the point; you're saying there is a possability the EMU DSP is not taking advantage of the headroom? Was it you who suggest the -70db? testtone to test this very concept on the EMU? I'd like to find out. Right now, digitizing even my hot 45's ...I could use a little more gain. My average RMS is about -20 prior to normalizing which I'd rather not even do, simply record hot. I could use the L2 compressor to tack on 3-5DBs but I'd rather not, I just want to record hot and do minimal noise reduction, otherwise I'll just raise the noise floor with normalize and live with it...compression feels to destructive for "archiving", I'm not mastering my own music.

v/r
DC