Hey,

I'm wondering whats the proper way to set up the volume for my music on my PC. There is a volume control on my amp, one on the control panel, one in my driver software, and one in my media player. With all these volume sliders, what is the right order to set them in? Should I set the amp loud, then set the ones on the PC low? Or the amp low and the PC high? I'm a little confused as to what is the right way. Thanks.

-Orbit

In general, digital domain volume controls reduce the bit depth. Taking a 16 bit CD as an example, lowering the volume digitally will give you a 15, or 14, or 12 bit output (depending on how much you lower the volume). This removes lower level detail. A volume control in the analogue domain is to be preferred.

So what you're saying is basically that we should put all the volume controls on the computer at maximum, and only use the volume control on ones speakers or amp for adjusting the volume? You're telling me that I actually lose fidelity and/or dynamics by not having my various computer volume settings at full blast?

I was under the impression that full sound output from the computer (or any sound source really) was bad, might result in distortion, etc? I always keep my computer volume at around 60%, and use my amp's volume control. This isn't right?

"Soft" Volume-Sliders (i.e. PC) should be set as high as reasonable. Then, the amplifier should be used to adjust the volume.

The reason for this only partially has to do with fidelity: the amp has the highest amplification-reserves - so, what you loose before the signal reaches the amp, will significantly reduce the amplification-potential of the whole signal-chain.

However, there is a reason why i wrote "as high as reasonable" and not "as high as possible"... in some cases you have to lower the signal a bit at the soft sliders:
- to have a "buffer" against clipping (applies when using EQs, DSPs, live-mixing, etc.)
- to average the loudness-level with differently mastered music (replaygain)
- your soundcard or its drivers suck if you move the sliders too high (applies to some crappy soundcards)



Nope. This may be the case with analogue devices(i.e. speakers) but not with digital devices. At least it shouldn't: there are some soundcards which behave like you described, but those can be considered "low quality". Usually, you should be able to set the volume at 100% and the result shall be a direct "pass-through" of the signal without altering it much. However, consider the exceptions mentioned earlier in this post.

Digital volume really is just simple maths: You've got a scale from 0.00 to 1.00. It cannot go above 1.00, else it will clip. Now asume we've got a source-signal at 0.90 peak(i.e. an mp3) and send it to the audioplayer... the volume-slider of the audioplayer then basically is just a multiplicator: if you set it to 80% then 0.90 x 0.8 = 0.72..... Okay, next the signal is send to the soundcard which you have set to 50% - that will do: 0.72 x 0.5 = 0.36!

What does the above example mean? Well, it means that if you send this signal to the amp, then the amp will only have 1/3 of its amplification-potential!

The SoundBlaster series, and possibly others from Creative, have built in limiting on recording. If you don't keep the input signal below -3dBfs it gets squashed. This limiting reduces fidelity and linearity in the upper range but also makes it hard to clip your recordings. Perhaps this is somewhat helpful to those who just want to proceed and don't care to learn how to get the best results.

Possibly these cards have some similar functioning on the output, which means the signal level going to the output jacks should also be kept below that level. Also, some cards have some sort of analogue amplifier on the output, they can drive headphone directly. This is almost certainly not a real HiFi amplifier and maybe it is easily driven into distortion. If so, then a compromise through lowered output signal might give a better net result in spite of the reduced bit depth and resulting loss of lower level detail.

While digital volume controls do operate by reducing the bit depth, 24 bit cards have some bits to spare for this purpose. The trade off there is that the upper bit range is always the most linear (SoundBlasters and possibly some other gaming cards aside). Therefore the best quality is obtained by utilizing the most significant bits. Thus it is still best not to turn the digital volume level down very far.

The previous post's comments about add-ons are true. Adjustments such as EQing can indeed raise peaks above 0dB. This isn't a problem within proper audio software where one would be mixing music in floating point format, but I'm not sure how it works in something like Winamp where you are playing a 16 bit integer file. If the digital volume control is applied before the EQ then it should be beneficial. If it is applied after the EQ it I don't see how it could do anything except just lower the level of everything -- again by reducing the bit depth. The peaks would already be clipped if the EQ amplification was enough to induce clipping and turning down the volume would not help.

Better soundcards usually have their own control panel software; they do not use the windows mixer. Some have hardware mixers on board that may be able to function intelligently in conjunction with their own volume controls to good advantage. The one on the Audiophile 2496, for instance, is a 36 bit DSP device which means you should be able to throw out a few bits at the bottom via the output level control and not effect fidelity. However, the default for this card, and the way I always use it, is to bypass the mixer chip.

Of course if you are using the sound card to directly drive a power amplifier, where there is no analogue level control, you aren't left with many other choices than on-the-computer volume control. Also, with the way much modern pop music is mastered, there probably isn't any real loss from throwing out half a dozen bits at the bottom.

Ok, I now have a better idea of how digital amplification works and how to use it properly. Thanks for clearing that up.

It's not really dependent on the sound card, more on the drivers. For example, with the Chaintech windows drivers the AV710 clips badly if the Wave slider is set over about 70% - at 100% it's outputting a square wave. The VIA drivers seem better, and don't clip with both sliders at 100%. I can attach oscilliscope photos if anybody is interested.

My cheap onboard sound doesn't clip at any volume setting, but my old (first-generation) SBLive used to clip at 100% Wave volume.

Uh, processing the DSP-phase(i.e. EQing) in 16bit integer and applying the volume-amp afterwards would almost be perverted - like "lets trash the signal and call the ambulance afterwards to increase the damage even more" - but it wouldn't surprise me if some audioplayers do this.

Well, I got a tad lost in all the talk here, but what everyone is saying is that, I should basically set the volume on my PC to 70% or so and then fine tune the volume with my amp?

My Creative Sound Blaster PCI 128 (rebadged Ensoniq) doesn't have any clipping with the outputs all at 100% with the latest drivers.

I'd just like to clarify this point a little. Andy states that the limiter makes it hard to clip your recordings. My experience is at odds with this. The nature of the limiter is not at all subtle. Basically, if you try and get recording levels above about -2dB (the exact level varies from card to card), what you get is extreme analogue saturation. The results are pretty much just as bad as digital clipping - the only difference being that, whereas with digital clipping you get the wavetops chopped off ruler-flat, from a Soundblaster you get the wavetops chopped off just as savagely but with a "ragged" top. To my ears the two results are equally objectionable.


I don't buy this. The only thing that this analogue clipping does is make metering difficult, since hard disk recording packages typically work out their metering from the digital samples that emerge from the soundcard. Since 0dB *never* emerges from a Soundblaster during recording, it's impossible to even suggest that clipping might be occuring. The average punter will end up wondering why their recordings sound so distorted when the peak record level never gets above -2dB.

I was only speculating as to why Creative might have taken such a perverted approach. But then, they have done so many perverted things, I suppose philosophy is just wasted effort.

Hmm sorry if I'm dragging this offtopic a bit, just want to understand exactly what is going on here. Are we talking about the actual analogue part of the card clipping here? ie, the line-in circuitry will clip 'loud' line-in signals?
I use my Audigy 2 ZS card a lot to record vinyl at 24 bit. I have the "Analogue mix to 0db (it goes up to +6db) and for LPs the Line-in level at about +3db (lowering it for shorter records that are pressed at a higher level). I never seem to have any problem with the card limiting, if I've got the line-level set too high it'll just hard clip at 0db (like anything not in floating point would). This is because the level from the record players pre-amp is relatively low and not pushing the input circuitry, turning the gains up in the mixer is a separate matter? Am I on the right wavelength here?

Yes, there is an analogue saturation/clipping problem somewhere in these cards. It isn't in the initial input buffer, since lowering the recording level via the soundcard drivers eliminates the clipping. It must be somewhere between the analogue gain stage and the A/D converter. You can easily see that it isn't digital clipping, from the distinctly "ragged" appearance of the clipped wavetops.


As far as I am aware, the Audigy range of cards don't suffer from the same problem. It seems particular to those cards that use the 137x chipset (originally acquired when Creative bought out Ensoniq) - ie. the PCI bus Soundblasters.

Thanks for that, set my mind at ease