A question about room reflections

[hemmick reef]

Reef,

First, let me say you asked a relevant and meaningful question. Real-world room acoustics and reverberation is a very complex issue to deal with, let alone understand. You did real-world research. Well done.

Take this example:

A choir sings in a large acoustically treated room. When they sing quietly, there appears to be little or no room effect. As they sing louder, not only does the effect of the acoustical environment become apparent, it changes in color, decay time, primary reflection, secondary reflections, etc. This change is based on the volume alone. What sounds glorious at full volume may sound lacking at quieter ones. This is difficult, if not impossible to achieve using artificial reverb. Typically, what you will notice when you send less (or more) signal to an artificial reverb is more or less of an effect that sounds very nearly the same.

Though more easily controlled and more flexible than a real room, reverb systems that do not do a very, very good job with room modelling do not emulate real spaces well, if at all, though they may very well produce a reverb that suits your purpose.

Also, when you record direct (or miced for that matter), do attempt to set levels that utilize the full resolution of the medium. That is, set your inputs such that the loudest signal from the instrument will not produce an over, but will come as close to 0 as is reasonable. There is no reason to leave room for cream, sugar, cinnamon, etc. as effects will be added during mixdown to another track.

People that don't take advantage of the recording medium are like people who crop digital pictures. The old saying goes "don't waste pixels". In photography, lost resolution is seen, in recording, it is heard. This is one more reason to record at 24 bit vs 16. More resolution, i.e. you can afford to lose a little during processing.

So, to the boneheads that said you asked a stupid question, I would say get out more and do a little research before calling someone else stupid.

Best wishes,

Poco

Thank you very much for your answer which makes complete sense to me.

The reason for asking my original question was because I noticed that, as you say, sending more or less signal to a synthetic reverb did not vary in the same way as a sound increased in a real room. I thought something different was happening with the reflections as the volume of my amp increases.

It may be comparable to the way velocity works with a drum or piano, I am not sure? But I imagine the reflections get more chaotic in a real room with increased volume which gives a different quality to a sound. I am not sure if synthetic software or hardware reverbs can replicate this? Maybe they have this parameter set at an optimum level?
Cheers

 

[Steve Henningsgard]

Part of what you probably won't be able to replicate comes from the fact that, no matter how much reverb you're adding, it's making calculations based on the guitar at the same volume: meaning, it won't be able to take into account the difference in sound you get with different amp volumes...

That wasn't a very well put together sentence now that I read it back hah

 

[Poco]

"I am not sure if synthetic software or hardware reverbs can replicate this? "

A room modeler comes the closest.

 

[snow lizard]

Also, when you record direct (or miced for that matter), do attempt to set levels that utilize the full resolution of the medium. That is, set your inputs such that the loudest signal from the instrument will not produce an over, but will come as close to 0 as is reasonable. There is no reason to leave room for cream, sugar, cinnamon, etc. as effects will be added during mixdown to another track.

People that don't take advantage of the recording medium are like people who crop digital pictures. The old saying goes "don't waste pixels". In photography, lost resolution is seen, in recording, it is heard. This is one more reason to record at 24 bit vs 16. More resolution, i.e. you can afford to lose a little during processing.

There's a lot of threads about this subject in here. About two a week or so.

Most people say you shouldn't print levels so hot that you're overdriving your input chain (preamp, converters, any effects that might be in the chain). The absolute scale of dBFS in digital is different than dBv, dBu, dBVU or dBSPL. Yes, they all work off of the decibel, but if your converters are calibrated so that 0 dBVU (RMS) is equal to -16 dBFS or so, (coulds be a bit higher or much lower) and if the transient content of the source pushes an extra 3 to 6 dB, if you push your levels to peak at 0 dBFS, you're overdriving your preamp by 10 dB.

Especially in 24 bit, it's not a great idea. Analog preamps don't clip at zero. They have headroom. Digital clips at zero, so you build the headroom into it by setting your levels properly, which is how it was designed to work. It has nothing to do with cameras.

People that have tried to print lower levels and ignore the "use all the bits" type of misinformation that's out there make the comment that their mixes come out much cleaner and easier.


sl

 

[brodgind]

There's a lot of threads about this subject in here. About two a week or so.

Most people say you shouldn't print levels so hot that you're overdriving your input chain (preamp, converters, any effects that might be in the chain). The absolute scale of dBFS in digital is different than dBv, dBu, dBVU or dBSPL. Yes, they all work off of the decibel, but if your converters are calibrated so that 0 dBVU (RMS) is equal to -16 dBFS or so, (coulds be a bit higher or much lower) and if the transient content of the source pushes an extra 3 to 6 dB, if you push your levels to peak at 0 dBFS, you're overdriving your preamp by 10 dB.

Especially in 24 bit, it's not a great idea. Analog preamps don't clip at zero. They have headroom. Digital clips at zero, so you build the headroom into it by setting your levels properly, which is how it was designed to work. It has nothing to do with cameras.

People that have tried to print lower levels and ignore the "use all the bits" type of misinformation that's out there make the comment that their mixes come out much cleaner and easier.


sl

I just read the manual for the hd24. It says "4. Have the performer play the loudest section of the song you're about to record. Adjust the gain controls of your mixer(trim, channel, and master) until the loudest notes fall just below the Clip indicator on the HD24's meter." So Alesis is saying use all the bit's?

 

[snow lizard]

I just read the manual for the hd24. It says "4. Have the performer play the loudest section of the song you're about to record. Adjust the gain controls of your mixer(trim, channel, and master) until the loudest notes fall just below the Clip indicator on the HD24's meter." So Alesis is saying use all the bit's?

You're on page 38.

On page 36:

To record-enable a track:

3. Adjust the levels on your mixer so that the "average" level is at -15 dB on the peak meters of the ADAT HD24 and the loudest section never goes beyond 0 dB.

Digital audio recording is different from analog recording, and therefore requires a different method when setting levels. For more information, see Setting the Recording Level on page 38.

Also, check out the section called "The meters" on page 37, and the two paragraphs below what you quoted.


sl

 

[BRIEFCASEMANX]

A cool experiment:

Copy the guitar recorded quietly to another track so you have two tracks of the exact same thing.

Add some distortion or an amp simulator plugin to the 2nd track. Now send this 2nd track to the reverb buss, but make it a PRE FADER send. Then mute this track. This will make the distorted track so that the dry signal is muted in the stereo buss, and the only thing you are hearing is reverb from the more distorted track. Don't send the less distorted track into the reverb at all.

This way you get the dry signal from the regular track, and the reverb from a more distorted track. You could also simply record 2 tracks of the same guitar part, but make one louder with the amp, and do the same thing, but without the distortion/amp simulator plug.

I've done something similar, where I had 2 takes of piano, and I used one completely dry, and used one only going to reverb. It was gave a cool and sort of creepy ambience.

 

[RAMI]

Now send this 2nd track to the reverb buss, but make it a PRE FADER send. Then mute this track. This will make the distorted track so that the dry signal is muted in the stereo buss, and the only thing you are hearing is reverb from the more distorted track.

Just be careful. On many systems, muting the track won't allow it to send the signal to the reverb. You might have to simply turn the fader all the way down....But, like BRIEFCASE said, make sure it's on "PRE-FADER" or you won't get anything.

 

[Poco]

Snow Lizard is suggesting that setting levels that approach 0 on the digital recorders meters is wrong. I disagree. Here is what the pertinent sections of the Alesis HD24 manual say (including the references that he made):

Alesis:
3. Adjust the levels on your mixer so that the
“average” level is at -15 dB on the peak meters
of the ADAT HD24 and the loudest section
never goes beyond 0 dB.

Me:
OK, that's fine. Average means that there will be as much above that point as below that point. Read this carefully before disagreeing.

Alesis:
Digital audio recording is different from analog
recording, and therefore requires a different method
when setting levels. For more information, see
Setting the Recording Level on page 38.

Setting the Recording Level (page 38)

Setting the correct recording levels is crucial to
making any recording sound its best. On any
digital recorder, the best resolution is found
when the maximum recording level of each track
falls just below the “Clip” point.

Me:
Still want to disagree? You will be at odds with the manufacturer. If your levels never approach zero, you are not taking advantage of the full resolution of the medium.

Alesis:
However, since
the HD24 is a 24-bit recorder, you don’t need to
push levels quite so hard to avoid noise and
distortion as you did in the past.

Me:
Ok, fine. I will not have to take the chance that the crazy jazz trumpet player will go from whisper to scream and create an over. That is, if I am recording at 24 bit versus 16 (as is well known), I have far more resolution, and I don't have to worry about not using it all. I can record a bit lower, thus a bit safer. That is all that means, period.

Alesis:
The Meters
Each of the tracks has its own 10-segment LCD
meter, with levels ranging from CLIP (0 dBFS) to
–60 dB. Levels within 6 dB (1 bit) of clipping are
shown in yellow. The standard nominal level for
ADAT, -15 dBFS, is two segments below that color
change. (The "-15" label is shown in red; this level
will usually equal "0 VU" on an analog mixer
connected to the input.)

Me:
So if I go over -15, I will overdrive the input stage of my analog mix-down mixer. That's good to know, except I don't use a mix-down mixer. I import my tracks into a DAW and mix digitally. This whole "don't get too close to zero" thing does not apply to anyone who uses a DAW to mix with, or a board with digital inputs. Nothing in the DAW complains when the levels dance at -6, -3, or -1. There are 65,536 gradations of volume in CD quality sound. USE THEM ALL. I guarantee you, that by the time Jo-Jo's Mastering House of Loud Music is done applying the Maximizer to it, it ain't no -15!

Poco

 

[RAMI]

So if I go over -15, I will overdrive the input stage of my analog mix-down mixer. That's good to know, except I don't use a mix-down mixer.

I'm not going to go point by point by point on what you said. Maybe someone else will. And I'm not saying I know more than the guy that wrote the Alesis manual...though I've read some manuals that make me question the "genius" of anyone writing them. But I'm no expert at this, so I can't dispute it.

But I'm wondering if, in the above statement, you might have it backwards. The way I undertand it is that if you're recording up around "0" (digital), the problem is not that you're going to distort your analog mixer at mix down time....
What I thought it means is that if you're recording into your DAW at 0 (digital), that means that somewhere along your recording path, you might have been running whatever analog gear you're using (pre-amp..etc...) too hot. So, if you're going in just under 0 (digital), that means something along your recording path is peaking at 18+ (analog)...or 15+, whatever the equivelant of 0 (digital) is this week. Or, you're overdriving your digital inputs and/or A/D converters. That's where the danger is. And since you're not losing anything by backing off, why not back off???

I hope I have that right. I'm not arguing with you because I'm not even sure if my above example is correct. It's just the way I understood it.

 

[iqi616]

Snow Lizard is suggesting that setting levels that approach 0 on the digital recorders meters is wrong. I disagree. Here is what the pertinent sections of the Alesis HD24 manual say (including the references that he made):

Alesis:
3. Adjust the levels on your mixer so that the
“average” level is at -15 dB on the peak meters
of the ADAT HD24 and the loudest section
never goes beyond 0 dB.

Me:
OK, that's fine. Average means that there will be as much above that point as below that point. Read this carefully before disagreeing.

Alesis:
Digital audio recording is different from analog
recording, and therefore requires a different method
when setting levels. For more information, see
Setting the Recording Level on page 38.

Setting the Recording Level (page 38)

Setting the correct recording levels is crucial to
making any recording sound its best. On any
digital recorder, the best resolution is found
when the maximum recording level of each track
falls just below the “Clip” point.

Me:
Still want to disagree? You will be at odds with the manufacturer. If your levels never approach zero, you are not taking advantage of the full resolution of the medium.

Alesis:
However, since
the HD24 is a 24-bit recorder, you don’t need to
push levels quite so hard to avoid noise and
distortion as you did in the past.

Me:
Ok, fine. I will not have to take the chance that the crazy jazz trumpet player will go from whisper to scream and create an over. That is, if I am recording at 24 bit versus 16 (as is well known), I have far more resolution, and I don't have to worry about not using it all. I can record a bit lower, thus a bit safer. That is all that means, period.

Alesis:
The Meters
Each of the tracks has its own 10-segment LCD
meter, with levels ranging from CLIP (0 dBFS) to
–60 dB. Levels within 6 dB (1 bit) of clipping are
shown in yellow. The standard nominal level for
ADAT, -15 dBFS, is two segments below that color
change. (The "-15" label is shown in red; this level
will usually equal "0 VU" on an analog mixer
connected to the input.)

Me:
So if I go over -15, I will overdrive the input stage of my analog mix-down mixer. That's good to know, except I don't use a mix-down mixer. I import my tracks into a DAW and mix digitally. This whole "don't get too close to zero" thing does not apply to anyone who uses a DAW to mix with, or a board with digital inputs. Nothing in the DAW complains when the levels dance at -6, -3, or -1. There are 65,536 gradations of volume in CD quality sound. USE THEM ALL. I guarantee you, that by the time Jo-Jo's Mastering House of Loud Music is done applying the Maximizer to it, it ain't no -15!

Poco

The bit you highlighted is an aside. They're still telling the user (in more than one place from your quotes) that the nominal input should be 0 VU = -15 dBFS on an analogue mixer (or whatever) connected to the input of the HD24.

Where does it say that you have to take advantage of the full resolution of the medium? You can choose not to. If you have an analogue mixing desk with +24 dB of headroom, do you ignore the meters and run it just short of clipping just because you can? Probably not.

Also, "average" does not at all mean that there will be just as much above as below that point - unless the dynamic range of the HD24 is 30 dB.

 

[Poco]

"The bit you highlighted is an aside. They're still telling the user (in more than one place from your quotes) that the nominal input should be 0 VU = -15 dBFS on an analogue mixer (or whatever) connected to the input of the HD24. Where does it say that you have to take advantage of the full resolution of the medium?"

With all due respect, this is getting a little ridiculous. What if the input device is not a mixer? What if I am using a POD, a mic pre, or whatever? These devices have plenty of output to drive my MX2424 meters into the red without distorting themselves (not that it won't, of course, create distortion in the recording device).

"They're still telling the user (in more than one place from your quotes) that the nominal input should be..."

No, they said the average input should be -15. Look up the difference between the words nominal and average, it may surprise you.

This is the real kicker:
"Where does it say that you have to take advantage of the full resolution of the medium?"

Where does it say you have to use the electric windows in your car? You could just open the door for some fresh air. Where does it say that you have to use all of the processor speed of your computer? You could lower the CPU voltage, and run it slower, with the added bonus that it would run cooler. You could do a lot of things that don't make sense if you want. It's your recording. Remember, if it doesn't make sense, it probably doesn't.

Greater sonic resolution is central to the idea of recording at 24 bit vs 16. Why up the resolution, then intentionally throw resolution away? I'm not talking about just using safer levels either. I'm talking about using half, as some would suggest. You want to see a pain in the neck? Try applying these low signals to compressors and effects. They don't work quite as you would expect when signals are low. You generally end up boosting the level through the gain control. Now there is a real waste. Record low, and resort to boosting gain. All commercial recordings end up at about the same average loudness. There are no low signals in the finished product. Why start out that way??

Try this: With a device that has sufficient output (i.e. one that does not create it's own distortion) record a sine wave @ -15, then record one @ -1. Play them both into an oscilloscope, and post photos of the display.

Later,

Poco

 

[RAMI]

" What if the input device is not a mixer?

I touched on that in my post. What difference does it make if it's a mixwe, a pre or any other A/D or analog gear?

 

[iqi616]

"The bit you highlighted is an aside. They're still telling the user (in more than one place from your quotes) that the nominal input should be 0 VU = -15 dBFS on an analogue mixer (or whatever) connected to the input of the HD24. Where does it say that you have to take advantage of the full resolution of the medium?"

With all due respect, this is getting a little ridiculous. What if the input device is not a mixer? What if I am using a POD, a mic pre, or whatever? These devices have plenty of output to drive my MX2424 meters into the red without distorting themselves (not that it won't, of course, create distortion in the recording device).

What does the "or whatever" in my statement mean other than a "POD, a mic pre, or whatever?"? See, even you said it.

"They're still telling the user (in more than one place from your quotes) that the nominal input should be..."

No, they said the average input should be -15. Look up the difference between the words nominal and average, it may surprise you.

I know the difference but that's not that important in the context of this discussion. When you're calibrating (you do calibrate your recording chain don't you?), you will be calibrating so the average signal level reads - 15 dB on the meter.

This is the real kicker:
"Where does it say that you have to take advantage of the full resolution of the medium?"

Where does it say you have to use the electric windows in your car? You could just open the door for some fresh air. Where does it say that you have to use all of the processor speed of your computer? You could lower the CPU voltage, and run it slower, with the added bonus that it would run cooler. You could do a lot of things that don't make sense if you want. It's your recording. Remember, if it doesn't make sense, it probably doesn't.

So recording at any level that does not take full advantage of the full resolution of the medium doesn't make sense? Tell that to a whole bunch of pros who don't run 2" 24-track at it's full resolution or their pro 24-bit gear at its full resolution. The word is "headroom". You have the extra quality so that you have the headroom so that can get on with the real job of recording music instead of getting all anal about whether you are using every possible bit of every possible word.

Greater sonic resolution is central to the idea of recording at 24 bit vs 16. Why up the resolution, then intentionally throw resolution away? I'm not talking about just using safer levels either. I'm talking about using half, as some would suggest. You want to see a pain in the neck? Try applying these low signals to compressors and effects. They don't work quite as you would expect when signals are low. You generally end up boosting the level through the gain control. Now there is a real waste. Record low, and resort to boosting gain. All commercial recordings end up at about the same average loudness. There are no low signals in the finished product. Why start out that way??

You don't record that low. -15 dBFS is not low, in fact it is 3 dB higher than the generally recommended level (which is 0 VU = -18 dBFS). -48 dBFS is low and I don't recall anyone recommending recording at that level. I guess that Alesis pick -15 dBFS because of their history with 16-bit ADAT. You can't leave too much headroom with 16-bit but you need to leave enough. It would be interesting to check out an original ADAT handbook and see if this is just a throwback that's been carried throughout their documentation.

You shouldn't spend all your life trying to stay close to clipping and throw away hundreds of great takes because you did clip (i.e. the engineer screwed up the recording). That is a far bigger waste than failing to make maximum use of the available bit-depth. To avoid losing takes you simply leave an adequate amount of headroom and concentrate on the music, not the science.

I wasted years agonizing over maximizing the signal level onto analogue tape or into my DACs. Now I just work with 18 dB headroom and almost never have to worry about clipping and I can concentrate on the most important part of the job - capturing the best performance. The intrinsic quality of 24-bit ensures that after multiple rounds of level changes and compression at different stages of the process, the 16-bit 44.1 kHz result has all the technical quality that it requires.

Try this: With a device that has sufficient output (i.e. one that does not create it's own distortion) record a sine wave @ -15, then record one @ -1. Play them both into an oscilloscope, and post photos of the display.

Later,

Poco

And what's that got to do with music? The important thing is whether your ears can hear the difference, not whether your eyes can see the difference.

 

[Poco]

Originally Posted by Poco
Try this: With a device that has sufficient output (i.e. one that does not create it's own distortion) record a sine wave @ -15, then record one @ -1. Play them both into an oscilloscope, and post photos of the display.

And what's that got to do with music? The important thing is whether your ears can hear the difference, not whether your eyes can see the difference.

The o'scope is the non-opinionated BS detector, that's what. A signal recorded at -1 will be just as sonically pure as one recorded at -15. If you think you can hear the difference of a particular signal recorded at a lower volume than the same signal recorded at a higher volume (assuming the dynamic variation is covered in its entirety) you're just fooling yourself. If you want to offer your clients less dynamics, that's your call. I prefer to give mine more.

The word is "headroom". You have the extra quality so that you have the headroom so that can get on with the real job of recording music instead of getting all anal about whether you are using every possible bit of every possible word..

First off, more headroom than is necessary to prevent clipping produces nothing in the way of "quality". Silence is silence. The bits don't get grungier as you approach 0.

Secondly, nowhere did I suggest that a reasonable amount of headroom (whatever is required to prevent a given performance from clipping) is undesirable, but saying that a given recording should never exceed -15, -12, or anything other than something just below 0 is just foolishness, and it is time to stop regurgitating foolishness. I don't know which 2" 24 track studios you've worked in, but the ones I remember just loved pushing levels. It's where we get the term "tape saturation".

I don't get "all anal" about squeezing every once of resolution I can into my recordings, but it would appear that you have gotten "all anal" about keeping it out.

Poco

 

[travelin travis]

I think Poco makes some good points. There are a few people here that tout lower levels as religion and I've been guilty of almost taking it as that. To be honest, I don't hear a difference in near zero levels and -18 average levels. I can understand the headroom argument but really, what's wrong with pulling down the faders after the signal is recorded? I agree with Poco on the point that plugins like to see a healthy signal. Some plugins react differently depending on the signal level feeding it. I was told in another thread that running levels near 0 is "cooking the converters". I don't buy it. I've compared some tracks at near 0 and -18 average and I don't hear any cooking, distortion, clipping, etc. Others have claimed that their mixes have improved since they began tracking at lower levels. Now surely some mic preamps or other gear will have a maximum level at which it's noise begins to affect the recorded signal. Maybe that's what is going on here for those people. My opinion is that you should evaluate your own gear and figure out what works best for you and not just take popular rumor as religion.

If anyone really wants to prove his/her point, post some examples of near 0 and -18 average tracks for comparison. Everyone here has the means to do so. Quit talking about it and prove it.

 

[RAMI]

The reason it's not good.....and this has been mentioned about 150 times, yet it's the one thing that the people still stuck in 16 bit recording techiniques don't seem to want to address.

There is nothing wrong with, and there is no difference in sound between staying at -18 or going up close to 0.

But for the 1000th time, this is the problem: If you're going into your DAW or digital recorder up around 0 digital, that menas you're over-driving something else along your signal path. And it's probably not giving you optimum sound because of it. O digital is like 15+ or 18+ analog. It's not smart to run that hot because it means that something in your signal chain is running too hot.

But some people would rather just avoid the real reason and keep arguing for the sake of arguing.

 

[iqi616]

Originally Posted by Poco
Try this: With a device that has sufficient output (i.e. one that does not create it's own distortion) record a sine wave @ -15, then record one @ -1. Play them both into an oscilloscope, and post photos of the display.




The o'scope is the non-opinionated BS detector, that's what. A signal recorded at -1 will be just as sonically pure as one recorded at -15. If you think you can hear the difference of a particular signal recorded at a lower volume than the same signal recorded at a higher volume (assuming the dynamic variation is covered in its entirety) you're just fooling yourself. If you want to offer your clients less dynamics, that's your call. I prefer to give mine more.



First off, more headroom than is necessary to prevent clipping produces nothing in the way of "quality". Silence is silence. The bits don't get grungier as you approach 0.

Secondly, nowhere did I suggest that a reasonable amount of headroom (whatever is required to prevent a given performance from clipping) is undesirable, but saying that a given recording should never exceed -15, -12, or anything other than something just below 0 is just foolishness, and it is time to stop regurgitating foolishness. I don't know which 2" 24 track studios you've worked in, but the ones I remember just loved pushing levels. It's where we get the term "tape saturation".

I don't get "all anal" about squeezing every once of resolution I can into my recordings, but it would appear that you have gotten "all anal" about keeping it out.

Poco

I am anal about avoiding losing takes due to clipping but I'm not anal about keeping out resolution nor do I say the signal should never exceed a particular level (other than 0 dBFS). I simply do not agree that the correct way to set a recording level is to have someone play loud and set the trim to just below clipping because in practice they almost always play louder in the actual take to the point where clipping is a serious risk. That is not the performer's problem - they are not the engineer. If the recording clips, it is the engineers fault. Losing takes due to clipping interferes with the performers confindence in your ability to capture them at their best.

I set a sensible amount of headroom (my definition of headroom being the difference between the averaged signal level and 0 dBFS) and trust the gear to deliver the quality without clipping while I concentrate on the more important task of working on the performance. Fussing over the odd few dBs to maximize a quality that already exceeds reasonable expectations is a distraction and life is so much easier since I stopped wasting my time worrying about it.

 

[iqi616]

I think Poco makes some good points. There are a few people here that tout lower levels as religion and I've been guilty of almost taking it as that. To be honest, I don't hear a difference in near zero levels and -18 average levels. I can understand the headroom argument but really, what's wrong with pulling down the faders after the signal is recorded? I agree with Poco on the point that plugins like to see a healthy signal. Some plugins react differently depending on the signal level feeding it. I was told in another thread that running levels near 0 is "cooking the converters". I don't buy it. I've compared some tracks at near 0 and -18 average and I don't hear any cooking, distortion, clipping, etc. Others have claimed that their mixes have improved since they began tracking at lower levels. Now surely some mic preamps or other gear will have a maximum level at which it's noise begins to affect the recorded signal. Maybe that's what is going on here for those people. My opinion is that you should evaluate your own gear and figure out what works best for you and not just take popular rumor as religion.

If anyone really wants to prove his/her point, post some examples of near 0 and -18 average tracks for comparison. Everyone here has the means to do so. Quit talking about it and prove it.

As poco is the one hung up on the bits and bytes I'll let him do the science to prove that setting the recording level so the averaged signal is around -18 dBFS will produce an audible loss of quality compared to a higher level.

I don't think there is an audible difference which is why stressing over maximizing recording levels is a distraction from a more important aspect of the job - capturing the best performance. The point I am making is that you should hit record and be 99.9% confident that you won't get a clipped signal without you having to watch the meters. Setting a sensibly safe headroom achieves that without any appreciable loss of quality.

 

[SouthSIDE Glen]

If you think you can hear the difference of a particular signal recorded at a lower volume than the same signal recorded at a higher volume (assuming the dynamic variation is covered in its entirety) you're just fooling yourself.

The reason those of who are saying record at an RMS somewhere in the high negative teens are saying that is not because of any questions involving digital resolution or that kind of thing. It's because that signal level coming out of the digital side of the converter represents a line level signal going into the analog side of the converter. 0dBVU analog =~ -20 to -12dBFS digital (depending on your converter's exact calibration.) It's all an issue of gain structure and staging.

With the exception of those engineers who use a special technique for overdriving the converter in special circumstances (don't get started on that here, it's besides the point), the point is, that if you're recording hotter than that on the digital side, that means one of two things is happening; either you're driving the gain hard on the analog input of the converter, or your pushing the bits up on the digital side after conversion. The point is, neither one of those helps your signal quality, and can, in fact degrade it.

First, on the analog side of the converter, just like with any other analog preamp, if you push the signal too hot, you get distortions; too cold and your S/N ratio suffers. Which is why somewhere around 0VU - usually calibrated to a line level of +4dBu - is the designed sweet spot for the main energy of the signal. Run into the converter at that level, and it will BY DESIGN come out the other side at an RMS of somewhere around -18dBFS (give or take), with that extra 18dB or so of headroom left over for crest factor. It's not that there's anything magical sounding about that digital level; it's that that digital level represents the sweet spot surrounding line level voltage (0VU) on the analog side.

But how about pushing the levels up on the digital side, post-converter? Why waste all those extra bits, you should use them all, right? Not really. Just as -16dBFS does not sound "better" than, say, -6dBFS - bits are bits after all, like Poco says - neither will pushing it to -6dBFS on the digital side sound any better than -16dBFS. In fact, you can make things sound worse when you do that. Not because of bit usage in and of itself, but because when you raise the gain of the signal, you're also raising the gain of the noise within the signal. Any noise floor coming from the analog side will be raised to higher digital levels by boosting the post-conversion digital gain. And to do so without adding any benefit to the signal itself means by definition a degradation in signal quality.

So, keep the gain staged somewhere around 0VU on the analog side of the converter to get optimum sound, let it fall out the digital side at unity gain, and you'll be getting the best signal staging you can. That signal will just so happen to RMS out in the mid to high negative teens FS on the digital scale because that's how the engineers designed it. And once it's in the digital scale, any further bit pushing will not help anything and can in fact serve to increase the noise level.

That's pretty much all there is to it. What throws a lot of people off is they never bother to differentiate between analog and digital measurements. They just say "0dB" whether they mean 0dBVU, 0dBu, or 0dBFS. The problem is, those are three entirely different values, and that leads to a lot of the misunderstanding.

G.