Hi folks. I'm currently reading the book "mixing with your mind", which was recommended some times ago in SOS. Very interesting, but I'm puzzled by the opinion of the author about digital recording. According to him, digital has intrinsically 0dB of headroom, the best signal is only at maximum resolution at (or just below) 0dBFS. So, to get the best signal when recording, you should "watch your transients" and hit the recorder as hot as possible. I actually thought that, especially at 24 bits, it was perfectly OK to leave a quite large headroom. According to the author again, the lower the signal is, the more grainy it is, and the more detail are lost, leaving only the loud part of the sound crystal clear. Has anyone comments about this, and has anybody else read the book?
I'd also like to have more info about the classification of the mics and sources according to their "hardness".
Cheers,
Eric

--------------------
dB Studio - recording, mixing, transfer
www.dBStudio-Liege.net

As I am aware, the louder your signal, the more bits it takes up, in effect, the more words your computer has used to describe it. Which could be what he means by detail. Also, the louder the signal, the better the signal to noise ratio on whatever analogue gear you used to record. So yeah, it's still a good idea to try and record as "hot" as possible

It's NOT a good idea to record very hot into a digital system...usually.

In terms of making the best use of available bits then on paper, the hotter the better, but in practice the cons outweigh the pros:

Say you leave ten dbs of clear headroom above your highest peak...A 24 bit system has a theoretical dynamic range of 144db. In practice lets say it's a hundred. That's still just the head on top of your pint. In addition people get confused about what 'resolution' means in this context. It is not the same as reducing sample rate.

The problems with leaving no headroom are rather more tangible:

1. You're going to get overloads.

2. You have no scope for additive processes like eq boosts without reducing gain within the system...which is pointless and has it's own disadvantages sonically.

3. Plug-ins that are designed to operate on -18dbfs signals can start to sound wierd and need to have their input levels pulled down all the time.

4. If your system is calibrated so that -18dbfs=0dbvu you're going to be running your pre's at +18. High end gear copes easily enough with this but it's not necessarily the sweet spot for the pre. Most analogue gear is designed to work optimally around 0vu. Where it really sounds best is a matter of experimentation but the optimum level of the pre should be considered more important than getting a hot level to digital.

5. This works in reverse...If you send a +18dbvu level to a hardware compressor you might find that its hard to set a threshold to just touch the peaks. You might find that boosting an eq causes analogue overloads. etc

All in all, there are no advantages, and quite a few disadvantages. I find that my highest peaks are somewhere between -10 and -4 most of the time. They might be lower.

There's one obvious exception...When printing mixes back to digital from analogue, it's worth printing them hot. You can do this as you know exactly how high the mix will peak. If you're sending stuff to a mastering engineer leave them a couple of db's of headroom.

J

--------------------
www.jackruston.com

well, i've only been recording at 16bit so far, but i've always found the recording sounds best when i hit it as hot as possible, but it depends what i'm recording how hot i records it, if that makes sense. eg, if the recorder (aw16g) says my snare drum or kick are just distorting, it usually isn't a problem- i don't hear the distortion and it sounds better hotter. (i don't think there is any form of limiter on the inputs, but i might be wrong, i know the aw4416 does have). Vocals on the other hand, I compress externally, and definitely sound better having compressed them to record hotter.

I've got that book as well, and it's improved my recordings immeasurably, but i don't think he's necessarily very orthodox in his viewpoints. Then again, I've found that from a purely subjective piont of view in my own recordings he's right,
C

Edited by Soulgreed (08/04/06 10:17 PM)

Quote:

---

Most analogue gear is designed to work optimally around 0vu.

---

How kind of them I could say that the reverse is also true


The book in question is interesting/entertaining enough as far as it goes but I think Bob Katz's book is far more relevant, up to date, informative, useful and important as a reference and model upon which to base your recording/mixing/etc. practices.

i try to keep the signal peaks within -6dbfs and -12dbfs

as mentioned above, 24-bit has 144db of dynamic range, which is an incredible amount. you can hit 24-bit digital at -60dbfs (which falls below the entire dynamic range of vinyl) and still have roughly the same resolution of 16-bit (CD quality) audio. so as long as you're above -60dbfs (which is basically inaudible) then you're still doing better than CD quality. surely if -60dbfs is a good enough resolution for CDs then -12dbfs will be more than adequate !

i'd much rather have the headroom than risk a clip and destroy a take.

--------------------
Statick Audio

Quote DrBob:

---

I'd also like to have more info about the classification of the mics and sources according to their "hardness".
Cheers,
Eric

---

I have read Mixing with your Mind again and again and I think it is great.

What I understand he means when he talks about "hardness" is the harshness and abrasiveness of a sound. It is kind of a subjective personal thing to do with the way you perceive the sound.

In practice I use the idea to select microphones. I only have 2 - an SM58 and a Rode NT1 and the Rode will "hear" much more top end - it makes everything sound "crisper" than the SM58. A better mic one may think. However, some people have a voice that is sibiliant or harsh and cutting in character and the Rode can exaggerate these qualities to the point where they sound unpleasant. These voices sound a lot more pleasant with the SM58.

In Stavrou's terms the Rode is a harder mic than the SM58 and he suggests that you use a hard mic on a soft voice. Using a hard mic on a hard voice can make the result too hard.

Quote Statick:

---

i try to keep the signal peaks within -6dbfs and -12dbfs

as mentioned above, 24-bit has 144db of dynamic range, which is an incredible amount. you can hit 24-bit digital at -60dbfs (which falls below the entire dynamic range of vinyl) and still have roughly the same resolution of 16-bit (CD quality) audio. so as long as you're above -60dbfs (which is basically inaudible) then you're still doing better than CD quality. surely if -60dbfs is a good enough resolution for CDs then -12dbfs will be more than adequate !

i'd much rather have the headroom than risk a clip and destroy a take.

---

I agree with all this but would n't a high level as practical give the best signal to noise ratio?

I have also read Strav's book, and I think it is a great read with lots of great info. It gets much better the second time you have read it however!

I have also read Katz's book, and it is also a great read, but they approach things from entirely different perspectives. Katz is more technical, and Strav is more "touchy-feely" using common sense and your ears rather than specs and technical aspects.

Both points have their uses.

I am now recording nothing that peaks beyond -6 dB to leave myself a bit of headroom and I am getting great-sounding and full recording.

--------------------
James
"Never interrupt your enemy when he is making a mistake" ~Napoleon Bonaparte~

Quote DrBob:

---

I'm puzzled by the opinion of the author about digital recording. According to him, digital has intrinsically 0dB of headroom, the best signal is only at maximum resolution at (or just below) 0dBFS.

---

Depending on your point of view this is either:
a) NOT TRUE
or b) NO DIFFERENT FROM ANALOGUE.

Digital has a finite upper level limit which you can not exceed. Analogue does too. With digital the limit point (0dBfs) is very obvious and precise, while with analogue it is a slightly grey area of rapidly increasing distortion.

Top avoid running into the distortion zone with analogue systems, we allow a headroom. A typical console, for example, will operate with nominal levels of +4dBu, with a headroom of about 20dB above that.

In the early days of digital, we didn't operate with much headroom because the converters were barely able to achieve 16 bit resolution and 96dB signal to noise -- way inferior to analogue kit. Also, some very early system weren't properly dithered, which meant that distortion and artefacts increased dramatically at low levels.

All these things have gone away now. Digital systems are 24 bit (with 130dB dynamic range easily achieveable), and properly dithered throughout.

So, the sensible engineer will leave a sensible level of headroom, just like on analogue systems of old.

Hugh


--------------------
Technical Editor, Sound On Sound

Quote Soulgreed:

---

well, i've only been recording at 16bit so far, but i've always found the recording sounds best when i hit it as hot as possible

---

I used to think this way. I found that some things sound immediately gratifying in isolation, but become a serious pain to mix later. Excercising restraint with compression (or omitting it altogether) from the recording chain (and recording an undistorted, fully dynamic signal) makes for material that's much clearer & easier to mix, IMO.

Quote:

---

i don't think there is any form of limiter on the inputs, but i might be wrong, i know the aw4416 does have

---

I'm afraid it doesn't. At all.

I configured a small 2-track system (ORTF mics, straight to DAT) to be used to record a rock-n-roll gig in my absence last night, for arrangement reference purposes. I calibrated it so that the average levels were about -20 or -18dBFS (approx. equal to 0VU in analogue terms, give or take a bit) and the highest peaks about -10dBFS.

Asides the obvious compromises of the basic set up & mic positioning, the recording is extremely clear & completely undistorted. Noise certainly isn't a problem, and this is at 16 bit!! I'd challenge anyone to tell me how recording hotter would have improved the recording, even momentarily ignoring the increased danger of clipping the DAT inputs.

Regards,

Mike

--------------------
www.ehsound.co.uk - Live Sound Hire & Services

Quote Michael Harrison:

---

Quote Soulgreed:

---

well, i've only been recording at 16bit so far, but i've always found the recording sounds best when i hit it as hot as possible

---

I used to think this way. I found that some things sound immediately gratifying in isolation, but become a serious pain to mix later. Excercising restraint with compression (or omitting it altogether) from the recording chain (and recording an undistorted, fully dynamic signal) makes for material that's much clearer & easier to mix, IMO.

Mike

---

aye, sorry, i meant to say that i only use compression on the way in on vocals, and for the style i do mostly (very heavy metal! heheh) i find it very difficult to get something usable without this. I'm probably recording most things with peaks of about -4dbFS at most (other tha the kick & snare i mentioned)

Quote Michael Harrison:

---

Quote:

---

i don't think there is any form of limiter on the inputs, but i might be wrong, i know the aw4416 does have

---

I'm afraid it doesn't. At all.

---

ah right, think am confusing the limiter shown in the circuit diagrams (as a safety feature- will this not influence the sound if you run it too hot?)


Just as an aside, is it more likely to help recording hotter if you have problems with a high noise floor in the studio?

cheers
C

Quote Soulgreed:

---

Just as an aside, is it more likely to help recording hotter if you have problems with a high noise floor in the studio?

cheers
C

---

The signal to noise ratio isn't related to recording levels. So, unless your're recording guitars in a noisy enviroment and you're able to crank up the amp and change the signal to noise ratio you're not getting any benefits.

Quote Soulgreed:

---

Quote Michael Harrison:

---

Quote:

---

i don't think there is any form of limiter on the inputs, but i might be wrong, i know the aw4416 does have

---

I'm afraid it doesn't. At all.

---

ah right, think am confusing the limiter shown in the circuit diagrams (as a safety feature- will this not influence the sound if you run it too hot?)

---

The built-in dynamics on the AW4416 are in the digital domain - meaning, if you record too hot & use compression, you're just compressing an already-clipped (at the A-D stage) signal.

To the best of my knowledge (been using this machine for some time) the analogue circuitry has no dynamic control at all. If it has, it's evaded my attention for the last 5 years!

Quote:

---

Just as an aside, is it more likely to help recording hotter if you have problems with a high noise floor in the studio?

---

If you mean in terms of acoustic noise, like traffic, or machine noise from a PC, or interference picked up by the instruments/audio inputs, then no.

Every dB of gain used to increase your signal level also applies to any noise already present. All you're achieving is a stronger recording of your noisefloor!

Mike

--------------------
www.ehsound.co.uk - Live Sound Hire & Services

Quote Michael Harrison:

---

I configured a small 2-track system (ORTF mics, straight to DAT) to be used to record a rock-n-roll gig in my absence last night, for arrangement reference purposes. I calibrated it so that the average levels were about -20 or -18dBFS (approx. equal to 0VU in analogue terms, give or take a bit) and the highest peaks about -10dBFS.

Asides the obvious compromises of the basic set up & mic positioning, the recording is extremely clear & completely undistorted. Noise certainly isn't a problem, and this is at 16 bit!! I'd challenge anyone to tell me how recording hotter would have improved the recording, even momentarily ignoring the increased danger of clipping the DAT inputs.

---

I totally agree. Recently I've had the chance to test a Presonus Firebox in a friends home studio and tracked a couple of acoustic instruments using an Oktava 219 at 24 bit. I deliberately used a nominal level of somewhat around -26 and -30dBFS and guess what... a perfectly acceptable, noiseless and open sounding recording. And the 219 ain't exactly super quiet.

Edited by archdake mkII (09/04/06 08:24 PM)

There are reasons to take care with recording levels(outboard analogue nominal levels being one), however to the best of my knowledge i really can't see how "headroom" within a 32-bit floating point environment has anything to do with it.

--------------------
Phatcat Studios--"I love the smell of Genelecs in the morning"

Hi, leave a good headroom margin to avoid overloads, I line up my mixer to read 0db on its output meters while simultaneously setting my master recorder to read -12db, thats a good compromise, it gives you enough level and good headroom. The higher the bit rate the better, but I work with 16 bit and I've never had any problems with these levels.

Take care,

Tony.

Hey Celcius,

Who mentioned 32bit floating?

Quote archdake mkII:

---

Hey Celcius,

Who mentioned 32bit floating?

---

I did.
I figured not everyone is mixing analogue or using a PT TDM rig, in which case they would(normally) be mixing/processing in a 32-bit floating point environment.(where headroom is not much of an issue AFAIK). Someone mentioned plugins and overloads which I assume relates to mixing.


--------------------
Phatcat Studios--"I love the smell of Genelecs in the morning"

Quote Celsius:

---

I figured not everyone is mixing analogue or using a PT TDM rig, in which case they would(normally) be mixing/processing in a 32-bit floating point environment.(where headroom is not much of an issue AFAIK). Someone mentioned plugins and overloads which I assume relates to mixing.

---

Headroom shouldn't be an issue with 32 bit domains... but not all systems are equally well engineered. Problems can and do arise, particularly when using third-party plug-ins that don't handle high level signals particualrly well.

The way to find out is to try running the system with sensible headrooms, even though it is a 32 bit domain, to avoid anything over 0dBfs anywhere in the chain. See if it sounds better than the same system running that much hotter. A lot of people have foudn that keepign levels down, and allowing sensible headrooms, results in an audible improvement in overall sound quality.

This isn't true of all systems, but it does seem to affect a surprising number, and especially when third-party plug-ins are involved as I said,

hugh

--------------------
Technical Editor, Sound On Sound

I totally agree with Hugh. On some 32bit f.p. platforms/DAWs I definately don't like the result of mixes that peaked above zero but are then brought down.

Quote archdake mkII:

---

I totally agree with Hugh. On some 32bit f.p. platforms/DAWs I definately don't like the result of mixes that peaked above zero but are then brought down.

---

You are ofcourse aware that the "0dB" limit on mixer channels operating with 32-bit fp is totally abstract.
Afaik it doesn't exceed anything and you certainly don't get any overflows.
Attenuating the master fader scales the number and in effect is equal to attenuating all the channel-faders.

What PLug/DAWS/Platforms btw?

--------------------
Phatcat Studios--"I love the smell of Genelecs in the morning"

Quote thefruitfarmer:

---

I agree with all this but would n't a high level as practical give the best signal to noise ratio?

---

Not at all - Digital by itself really has no noise to speak of. The analog gear in the front end does - The distortion, noise, loss of focus and clarity when driving a preamp's output stage that far above nominal makes MUCH more of a difference than the lousy loss of a couple of bits which were really designed to be headroom anyway for the most part. The S/N of the gear is designed to be optimal at 0dBVU - It's whacked out and silly approaching the clipping stage.

Remember - 24-bit wasn't celebrated because you could record louder - It was celebrated because you could record SO much QUIETER (not "quieter" - more like "normal" - where your gear was designed to run anyway).

Seriously - Don't take anyone's word for it - Go do it. I will all but guarantee it'll change the way you record (to the way you're supposed to in the first place).

--------------------
John Scrip - MASSIVE Mastering - BLOG

Quote MASSIVE Mastering:

---

Seriously - Don't take anyone's word for it - Go do it. I will all but guarantee it'll change the way you record (to the way you're supposed to in the first place).

---

I have and it did.

Quote Celsius:

---

Attenuating the master fader scales the number and in effect is equal to attenuating all the channel-faders.

---

Yes, that is the way it should work -- and it does on many systems. However, not all software is written as well as it should be, and sometimes what should happen doesn't. It is simple enough to experiment and see for yourself with your own equipment whether 0dBfs+ signals cause problems or not.

hugh

--------------------
Technical Editor, Sound On Sound

Quote Celsius:

---

What PLug/DAWS/Platforms btw?

---

DirectX springs to mind. it's crap!

BTW, the original question was about headroom in the A to D. Headroom after the A to D is not as easily defined, because different programmers implement the bits in different ways. Not all 32 bit float systems are created equal. For example, Cubase VST 32 bit fp sounds inferior to Cubase SX 32 bit fp.

Just making sure we're talking about the same stuff!

--------------------
madtheory creations
Synths and pianos for Kontakt

Quote MASSIVE Mastering:

---

Quote thefruitfarmer:

---

I agree with all this but would n't a high level as practical give the best signal to noise ratio?

---

Not at all - Digital by itself really has no noise to speak of. The analog gear in the front end does - The distortion, noise, loss of focus and clarity when driving a preamp's output stage that far above nominal makes MUCH more of a difference than the lousy loss of a couple of bits which were really designed to be headroom anyway for the most part. The S/N of the gear is designed to be optimal at 0dBVU - It's whacked out and silly approaching the clipping stage.

Remember - 24-bit wasn't celebrated because you could record louder - It was celebrated because you could record SO much QUIETER (not "quieter" - more like "normal" - where your gear was designed to run anyway).

Seriously - Don't take anyone's word for it - Go do it. I will all but guarantee it'll change the way you record (to the way you're supposed to in the first place).

---

Thanks Massive.

I will give that a go....

I was thinking "the more bits the better the detail" and I appreciate you highlighting what else is going on at the point just below clipping.

Perhaps going for signal peaks around -12dB to -6dB is going to give better quality.

There's a similar discussion on the Yamaha DTXtreme forum, where in my ignorance I banged heads with a Yamaha techie on this very issue . So if you want to be more enlightened and/or confused, follow the link.
http://www.dtxperience.com/forums/showflat.php?Cat=8&Board=DTXtremeIIS&Num ber=197993&page=&view=&sb=&o=&fpart=1&vc=1
By the way, I believe everything I read (in Sound on Sound by Hugh Robjohns). That's what makes me more of a selective human being.

--------------------
Battle flags are flown at the feet of a garden gnome.

Quote thefruitfarmer:

---

Quote MASSIVE Mastering:

---

Quote thefruitfarmer:

---

I agree with all this but would n't a high level as practical give the best signal to noise ratio?

---

Not at all - Digital by itself really has no noise to speak of. The analog gear in the front end does - The distortion, noise, loss of focus and clarity when driving a preamp's output stage that far above nominal makes MUCH more of a difference than the lousy loss of a couple of bits which were really designed to be headroom anyway for the most part. The S/N of the gear is designed to be optimal at 0dBVU - It's whacked out and silly approaching the clipping stage.

Remember - 24-bit wasn't celebrated because you could record louder - It was celebrated because you could record SO much QUIETER (not "quieter" - more like "normal" - where your gear was designed to run anyway).

Seriously - Don't take anyone's word for it - Go do it. I will all but guarantee it'll change the way you record (to the way you're supposed to in the first place).

---

Thanks Massive.

I will give that a go....

I was thinking "the more bits the better the detail" and I appreciate you highlighting what else is going on at the point just below clipping.

Perhaps going for signal peaks around -12dB to -6dB is going to give better quality.

---

It's a double-edged sword... In a perfect world, more bits = better detail. But the amount of detail you *lose* and the damage to the signal from the drastic invasion of the headroom is a MUCH bigger deal - That headroom is there to support transients - Not the entire signal. There are a lot of places to screw everything up. That's why I just put everything at unity, turn up the pre-gain til the meat is riding at 0dBVU and go. Automatic recording levels, and mixes that almost mix themselves - So little effort.

I had a client send in a project a few years ago where everything was pushing red. Oh my god... What a nightmare just getting everything down far enough that the mix wouldn't clip. Here I am turning down *every single track* somwhere around 15-20dB just to *start* a mix. Turning the noisy, distorted, unclear, unfocused tracks down.

Where if everything was recorded at around 0dBVU, I probably could've put everything up at unity and started tweaking straight away. Compress this or that with NO make-up gain most of the time, no worries about adding modulation effects or broad EQ strokes.

Anyway - When you think about that, the tracking engineer *amplified* everything - a lot - to get it that hot. We all know what happens when you turn the preamp on a guitar amp up - It distorts. Same thing with a mic preamp to a lesser extent.

Basically, the guy was adding a little bit of distortion to every single track (around 40 of them). Figure 25 or those tracks were up at any given time. That's a lot of crap to screw up the mix. You might as well throw packing blankets on the speakers.

Sidenote - I called the engineer about it. I had to have the band retrack some of the guitars. I said "Hey, can you keep it around 0dBVU this time?" He said "You mean those levels weren't hot enough?"

Commercial facility here by the way... And the guy had absolutely no idea what I was talking about.

Long story short - I still get cards every holiday from him. He sent this big letter on how it "changed his life" and yada, yada. He thought that there was some big secret to getting that "big, airy, clear pro sound" that he never got on any of his recordings. His entire clientel has changed since. He was always the "cheap demo" studio - Now it's more of the "great sounding studio on a budget."

--------------------
John Scrip - MASSIVE Mastering - BLOG

'That headroom is there to support transients - Not the entire signal'

I'm not sure I understand you... I don't think anyone was suggesting that 0dB should be the average level or anything: just that the odd transient should get near to it. Doesn't everyone already know that you can't go above 0dB with a digital recording? Nobody intentionally clips transients, do they? Should transient peaks approach 0dB or not? If not, why not? Because the digital mixing algorithms can't take it (why can't they)? Or as Hugh suggests, because 3rd party plug-in processors may be badly written and end up clipping a 'hot' signal? I'm prepared to take your word for it of course, but I'd like to know why I shouldn't be using the full dynamic range of my converters- (sensibly of course, allowing for transients). What would you suggest is a sensible PEAK level for digital recording (i.e. a level above which no transient should go)? To put it another way: if a signal doesn't clip, how is it 'too hot'?

--------------------
Battle flags are flown at the feet of a garden gnome.

Quote Richard Graham:

---

Nobody intentionally clips transients, do they? ?

Quote:

---

.... just about everybody in the Loudness Mastering War seems to do this (just have a look at the waveforms of most commercial pop records these days- all the tops chopped off...) Very good converters can often take a bit of headroom crunching but you don't want to try it with cheapo stuff- otherwise I take 0db as the ultimate limit but I'm happy for a couple of peaks to hit it during the course of a song.

In the end, if there's a problem, you should be able to HEAR it- use your ears, not just your eyes !

tomafd

--------------------
http://anotherfineday.bandcamp.com/ http://anotherfineday.co.uk http://apollomusic.co.uk

Quote Richard Graham:

---

'That headroom is there to support transients - Not the entire signal'

I'm not sure I understand you... I don't think anyone was suggesting that 0dB should be the average level or anything: just that the odd transient should get near to it. Doesn't everyone already know that you can't go above 0dB with a digital recording? Nobody intentionally clips transients, do they? Should transient peaks approach 0dB or not? If not, why not? Because the digital mixing algorithms can't take it (why can't they)? Or as Hugh suggests, because 3rd party plug-in processors may be badly written and end up clipping a 'hot' signal? I'm prepared to take your word for it of course, but I'd like to know why I shouldn't be using the full dynamic range of my converters- (sensibly of course, allowing for transients). What would you suggest is a sensible PEAK level for digital recording (i.e. a level above which no transient should go)? To put it another way: if a signal doesn't clip, how is it 'too hot'?

---

Richard, if the transients approach 0 dBFS, then after processing, the transients can easily clip. Processing (including EQ, compression, SRC, etc.) can raise the level of the signal. If your transients are approaching 0 dBFS, you are in the "danger zone for sure. Simple as that. I don't want *anything* to clip, including transients! I am mixing to -6dB on transients.

Quote tomafd:

---

Quote Richard Graham:

---

Nobody intentionally clips transients, do they? ?

---

.... just about everybody in the Loudness Mastering War seems to do this (just have a look at the waveforms of most commercial pop records these days- all the tops chopped off...)

---

A lot of those full scale flat waveforms have been brickwall limited, not necessarily clipped, but you are right, there are still tons of clips in most over-loud mixes nowadays. I just want to make sure that it is clear that flat-topped waveforms are not *necessarily* clipped.

--------------------
James
"Never interrupt your enemy when he is making a mistake" ~Napoleon Bonaparte~

>>>> just about everybody in the Loudness Mastering War seems to do this (just have a look at the waveforms of most commercial pop records these days- all the tops chopped off... <<<

I thought there was a major difference between limiting and clipping, and mastering people used limiting (which doesn't just square off the wave-tops but maximises the gain of the overall waveform using fast but non-zero attack and release parameters)

--------------------
Battle flags are flown at the feet of a garden gnome.

>>>if the transients approach 0 dBFS, then after processing, the transients can easily clip. Processing (including EQ, compression, SRC, etc.) can raise the level of the signal. If your transients are approaching 0 dBFS, you are in the "danger zone for sure. Simple as that. I don't want *anything* to clip, including transients! I am mixing to -6dB on transients<<<


I take your point, that the damage is done by the effects that we pass the 'hot' signal through, rather than the signal being intrinsically damaged. So wouldn't it be possible (and even sensible) for software processors to include a front-end gain/trim control and/or at least a 'clipping' indicator- the way that hardware processors often do? Failing that, couldn't I ensure that clipping doesn't occur in EQ, by only cutting, and not boosting (I know that some EQs sometimes boost at the frequencies around the notch, so this wouldn't work 100% of the time, but still)! Isn't it good EQ technique in any case, cutting rather than boosting?

--------------------
Battle flags are flown at the feet of a garden gnome.

Quote Richard Graham:

---

...Nobody intentionally clips transients, do they?...

---

I do! Sometimes. For creative purposes.

That doesn't surprise me as I've heard some Dutch music. It wasn't a pleasant experience

Quote Richard Graham:

---

>>>> just about everybody in the Loudness Mastering War seems to do this (just have a look at the waveforms of most commercial pop records these days- all the tops chopped off... <<<

I thought there was a major difference between limiting and clipping, and mastering people used limiting (which doesn't just square off the wave-tops but maximises the gain of the overall waveform using fast but non-zero attack and release parameters)

---

Good discussion!!!

You are right in there is a major difference between limiting and clipping.

My comment was just that a flattened waveform could be due to excessive limiting and still not clip. A flattened waveform could be limited so that *nothing* goes above -6 dB, including transients (although when squashed to death, there is not such thing as a transient!), but the average levels of everything are pulled up to that same level.

In this example, you would get a flat waveform, but no clipping.

In real life however, the whole point of limiting masters to death is to squeeze as much volume as you can out of the song, so most of these flattened waveforms that you see are right up there at 0 dBFS, so there is indeed a fair amount of clipping.

--------------------
James
"Never interrupt your enemy when he is making a mistake" ~Napoleon Bonaparte~

Quote Richard Graham:

---

>>>if the transients approach 0 dBFS, then after processing, the transients can easily clip. Processing (including EQ, compression, SRC, etc.) can raise the level of the signal. If your transients are approaching 0 dBFS, you are in the "danger zone for sure. Simple as that. I don't want *anything* to clip, including transients! I am mixing to -6dB on transients<<<


I take your point, that the damage is done by the effects that we pass the 'hot' signal through, rather than the signal being intrinsically damaged. So wouldn't it be possible (and even sensible) for software processors to include a front-end gain/trim control and/or at least a 'clipping' indicator- the way that hardware processors often do? Failing that, couldn't I ensure that clipping doesn't occur in EQ, by only cutting, and not boosting (I know that some EQs sometimes boost at the frequencies around the notch, so this wouldn't work 100% of the time, but still)! Isn't it good EQ technique in any case, cutting rather than boosting?

---

Well...not quite, but you are certainly on the right track!

Any EQ boosts (as you mention) will increase the level, but also other things during the processing will do so as well, even if you do not boost any frequencies. SRC is one major culpret that will raise the level slightly during its processing process (there are several conversions going on under the bonnet).

I am getting in over my head now, so hopefully someone with more techinical knowledge than I will jump in here and explain *why* this happens. I just know that it does, not necessarily what is going on inside of the algorythm.

--------------------
James
"Never interrupt your enemy when he is making a mistake" ~Napoleon Bonaparte~

Thanks. What's SRC when it's at home then? (sorry just done a search: sample rate conversion ?)

i always keep my peaks below -6db at all times, for the reasons mentioned above, and also because some sound sources can be rather unpredictable, and even when you've checked the levels a dozen times there's always gonna be one snare hit / guitar strum that jumps out. when budgets are tight and time is a pressure, you don't want to be the one calling another take because something clipped.

--------------------
Statick Audio