DAW Life - To Pad or Not To Pad.

[KeyboardEric]

I was doing some recording of keys into my DAW this evening and I noticed the -20db pad buttons on the back of my audio interface. Wasn't sure what to do so I just pushed them in to be safe before laying down tracks. Things seem to be sounding good this way. Just wondering about what other folks do with this when recording keyboards into a DAW. cheers!

[Mark_OA]

I watch my levels and try to record to about -18 dB, riding the faders in my DAW to compensate and accentuate when needed afterwards.

[KeyboardEric]

me too!

[BbAltered]

When it is just me recording myself, there are no signals so hot that -20dB pad is needed. When recording others, I have on occasion used the -20dB pad to tame a hot guitar or drum signal.

[Stokely]

I don't record keyboards (I use softsynths) but when recording vocals/guitars my old habit is to try to get close to zero without distortion. It has been explained to me that this isn't as big a deal as it used to be (when I learned, we were using analog gear to analog tape) -- in those days sometimes you actually might WANT to "go over" and get a bit of saturation. With digital, no way, can't go over, and you can record at lower gain and it's not as critical. I still try for what I think is optimal gain structure to reduce noise.

 

So I'd play around with trying to get the best signal-to-noise out of the chain--perhaps instead of putting a pad on the interface, you'd get better results out of reducing the keyboard master (I've noticed on mine that turning them way up can be noisy). OR, it could be worse than what you have I've even read recently about how some synths sound better with patches that are turned down at the oscillator mix level (granted, it was talking about analog and iirc it was the Prophet rev2 they were discussing) to get a smoother sound. Basically if you are getting a noise-free signal without the pad then that's great, it just makes sure you won't distort.

[Docbop]

I try to avoid using pads because they tend muffle the sound.

[KuruPrionz]

Gain staging can be broken down into simple steps.

 

Plug a set of good quality headphones into your interface, play some music and get your volume on your headphone amp set to a reasonable level, not too quiet and not blaring loud. Then set it up for monitoring from the channel in your DAW you are going to record to, a "round trip" situation.

Turn the keyboard all the way down, plugged into your chosen channel on the interface.

 

Turn the gain control (knob on the interface or a software control, either one) up until you start hearing any hiss, hum or other noise. Then turn it down until those sounds disappear again.

Now start playing a low note on your keyboard, strike it with some vigor, we want a loud note. With your other hand, slowly turn the volume knob on the keyboard until you see your loud note bouncing the meter on the DAW at around -6db. If you get to close to zero some of your transients may be going into distortion but the meter is not fast enough to show that. Most modern interfaces have some overload protection but it will still sound better if you stay below that threshold.

 

If you don't hear noise again at -6db then you are good to go. If there is noise and you can make it go away by turning your keyboard down a little, do that. If you have to turn things down a lot to get rid of noise, something is wrong. A power conditioner/noise eliminator power supply can make a big difference. Sometimes it is as simple as a cord or a jack needs to be replaced. Try to find it and kill it!

 

If you can't, just record. Depending on how you mix the final product it may not be too noisy.

 

Same principle apply with anything you are recording. Set up your recording from back to front, keeping noise at bay.

[KeyboardEric]

Thanks for the tips! I've been gain staging everything at around -18db even with the -20db pad depressed. There is no noise that I can tell or any issues with this approach so far. My question was one of curiosity of how others may approach this option.

 

There doesn't seem to be a consensus in this thread on whether or not the pad should be depressed when recording keyboards. Maybe it depends on the keyboard and whether the signal is analog or digital. I might just go back and retrack some things with it un-depressed to see if there is any difference at all in the sound quality.

[Steve Nathan]

I use my ears, and have always subscribed to the mantra "If it sounds good...it is good"

[KuruPrionz]

I use my ears, and have always subscribed to the mantra "If it sounds good...it is good"

 

 

This is universally true.

At the same time, we've been given easy to use tools to maximise signal to noise ratio.

 

This is a good thing. Sounds good? What if you could sound better still? Cheers, Kuru

[Dr Mike Metlay]

As others have said, this is completely dependent on application and source. For my part, I avoid using pads unless my inputs are getting clobbered by way-too-hot signals that I can't easily turn down before they get to me. I don't tend to use them as gain-staging tools, but more as disaster-preventive tools.

 

[EricBarker]

I try to go pretty hot, but no where near clipping. I don't really have a set level, maybe absolute PEAK at -6. But it really depends. Not a terrible strategy with modern interfaces. These days converter noise floor is better than ever (contrary to Mr Neal Young), so 18 would probably be okay. But I'd be a little weary of going too far. Safety's good, clarity's good too. But as long as you're not peaking, nothing's wrong with being hot. Most of my tracks get dropped by a good 9 to 15 dB in the end.

[analogika]

I use my ears, and have always subscribed to the mantra "If it sounds good...it is good"

 

 

This is universally true.

At the same time, we've been given easy to use tools to maximise signal to noise ratio.

 

This is a good thing. Sounds good? What if you could sound better still? Cheers, Kuru

Exactly.

 

"If it sounds good, it is good" is a great maxim until you discover that it would have been so easy to make it sound so much better all this time.

 

Can't unhear "better".

[BbAltered]

I use my ears, and have always subscribed to the mantra "If it sounds good...it is good"

 

This works well for live music. For DAWs operating with a 64 floating point mix engine, the distortion of a hot signal is hidden from you until you render it as 16-bit CD audio (or mp3), when it becomes apparent as obnoxious digital clipping - i.e. even tho' it sounds good, it's shit.

 

My mixer (Mackie) is my AD/DA converter. I use the mixing board pre-amp trims to ensure an appropriate level at the mixing board, and the signal is robust but never clipped in the DAW.

[Reezekeys]

This works well for live music. For DAWs operating with a 64 floating point mix engine, the distortion of a hot signal is hidden from you until you render it as 16-bit CD audio (or mp3), when it becomes apparent as obnoxious digital clipping - i.e. even tho' it sounds good, it's shit.

Are you sure about this? I believe that having a 64-bit mix engine in a DAW has little to do in preventing issues of overloading the input stage of an analog console or a-to-d converter (and thus can't "hide" them), which to me is what a pad is for. A 64 bit mix engine is for all the math going on inside a DAW as you pass the signal through mix buses, effects, limiters, maximisers, etc. If the mix sounds good in a DAW, putting it onto a 16-bit CD shouldn't suddenly reveal any "shit."

[BbAltered]

I believe that having a 64-bit mix engine in a DAW has little to do in preventing issues of overloading the input stage of an analog console or a-to-d converter (and thus can't "hide" them), which to me is what a pad is for. A 64 bit mix engine is for all the math going on inside a DAW as you pass the signal through mix buses, effects, limiters, maximisers, etc. If the mix sounds good in a DAW, putting it onto a 16-bit CD shouldn't suddenly reveal any "shit."

 

Exactly correct: if one records distortion and noise (whether using a -20dB pad or not), then one will have distortion and noise at playback. However, with modern DAWs, you can record a distortion- and noise-free signal, and add a lot of EQ so that the track is clipping, and never hear that clipping until you render the track to CD audio. Hence: the old adage "if it sounds good, it is good" DOES NOT APPLY in the world of DAW recording and playback.

 

[Reezekeys]

I think I know what you're saying â using a DAW with a 64-bit floating point mix engine lets you apply extreme EQ and gain without clipping internally â but as you apply these extreme changes, aren't you still monitoring through whatever final bit-depth your D-to-A is? You'll hear any clipping before you render it to CD, no?

 

A few posts up you said "even tho' it sounds good, it's shit." How can something that "sounds good" be "shit"?

[KuruPrionz]

I believe that having a 64-bit mix engine in a DAW has little to do in preventing issues of overloading the input stage of an analog console or a-to-d converter (and thus can't "hide" them), which to me is what a pad is for. A 64 bit mix engine is for all the math going on inside a DAW as you pass the signal through mix buses, effects, limiters, maximisers, etc. If the mix sounds good in a DAW, putting it onto a 16-bit CD shouldn't suddenly reveal any "shit."

 

Exactly correct: if one records distortion and noise (whether using a -20dB pad or not), then one will have distortion and noise at playback. However, with modern DAWs, you can record a distortion- and noise-free signal, and add a lot of EQ so that the track is clipping, and never hear that clipping until you render the track to CD audio. Hence: the old adage "if it sounds good, it is good" DOES NOT APPLY in the world of DAW recording and playback.

 

We are probably off-topic at this point. So it goes...

 

It is wise to capture the best possible signal at the initial capture. The lower your signal to noise ratio, the better. You DO want to stay safe. I would not record a snare drum at -6, more like -12. Keys do not have anything close to the transient attack of a smartly struck snare drum so -6 is pretty safe for keys.

 

Before I mix, I strive to get the tones right in each individual track. I leave my channel faders at 0db when I do this. At the recording stage (A to D), one could pre-EQ and use either boost or cut with the volume at the output of the chain adjusted accordingly.

 

Once you are in the DAW, boosting must be done carefully and selectively.

 

I have seen channels that solo perfectly fine on both the channel meter AND the output section meter but when you play multiple channels together (with their channel faders set to 0), the sum of those channels will overdrive the output meter. I can hear it and see it.

 

It is usually low frequency "bloat" causing the problem. There really only needs to be one or two tracks that contain noticeable low end, the rest just causes a mucky sounding mix. This is where subtractive EQ is your friend in the context of an overall mix. Bloated low frequencies do not sound good and are not good.

 

It is easier and sounds better to "start high and work down" than it is to "play safe and need to turn up" since you are not sacrificing signal to noise when you turn down, if there was any noise it will vanish.

If you recorded "safe" and turn up, you may find that your track has more noise than you realized, it was simply hidden by lower volume.

 

If you mix your volumes before you mix your tones, I could see how it is possible to drive a track into distortion. I am not sure why you don't hear it? Playback system? Track is down too low for the distortion to be audible?

 

FWIW, I am also big on making duplicates of tracks, keeping the original track as recorded and putting plugins on these duplicate tracks. Then I use automation to blend the tracks as needed to serve the song.

 

So far, so good. Cheers, Kuru

 

[BbAltered]

I think I know what you're saying â using a DAW with a 64-bit floating point mix engine lets you apply extreme EQ and gain without clipping internally â but as you apply these extreme changes, aren't you still monitoring through whatever final bit-depth your D-to-A is? You'll hear any clipping before you render it to CD, no?

 

A few posts up you said "even tho' it sounds good, it's shit." How can something that "sounds good" be "shit"?

 

Try it some time with your DAW: record clean quiet audio into your DAW and add gain until the main bus meter goes into the red. When simply monitoring, the audio sounds clean. When rendered to CD audio, you will hear a lot of digital distortion.

[Jim Alfredson]

In the digital world, your signal should be peaking at around -12dbfs. At 24bit how you get there is pretty inconsequential in regards to noise.

 

Regarding internal mix busses, you cannot overload modern DAW mix busses, especially if they are running at 64bit. The overload comes from the digital to analog conversion. See the video below for proof.

 

 

[Reezekeys]

In the digital world, your signal should be peaking at around -12dbfs. At 24bit how you get there is pretty inconsequential in regards to noise.

 

Regarding internal mix busses, you cannot overload modern DAW mix busses, especially if they are running at 64bit. The overload comes from the digital to analog conversion. See the video below for proof.

 

Thanks for this post, I was starting to feel like Jack Nicholson in One Flew Over the Cuckoo's Nest here.

 

Recording to analog tape, you do want to get near 0dbu to stay above the noise floor. If you occasionally peak over 0db it's not armageddon, you're just saturating the tape a bit and adding that "warm" and "phat" (pick your adjective) "analog sound" (is this still a thing?). Digital is of course quite different; go above 0db in the converter and if it's more than a hundred samples in a row (or less, I never tested), your ears will let you know. The engineers I know do exactly as Jim says, record with the meters around -12.

 

And sorry BbAltered, I tried the experiment you suggested and my results mirror 100% the video Jim linked to. Bumping the gain up on an audio channel turned things to digital mush as I monitored. Turning down the master to compensate brought back undistorted audio, proving that an internal 64-bit mix bus has ample headroom to avoid digital clipping. To be sure, I saved a CD disk image file (I don't have any actual CD blanks to burn) and played the resulting AIFF file â clean as a whistle. There was no way that things sounded good while monitoring but suddenly turned to crap when rendered to a CD.

[EricBarker]

Hypothetically... if you NEVER get a digital over, is there ANY reason to record anywhere below -0.5dB? Seriously asking here, because this is kind of advanced digital acoustic physics. From a mathematical standpoint, I can't think of any reason why anything below clipping could ever be bad for tracking, in theory. HOWEVER, I've heard people swear rounding errors from hearty attenuation causing distortion. This seems extremely likely in a float environment.

 

I think there are obvious practical reasons for keeping levels conservative during tracking sessions. Everyone has their rule of thumb, but it's hard to believe that there is an wholistic, Einsteinian universal constant for which we all should obey. For instance: driven guitars naturally compress, so you can roll fairly hot with axes and not worry so much. Conversely, vocalists can surprise the hell out of you when they get into it, so best to err on the side of caution.

 

My rule with modern digital audio: Set it so you could never possibly clip from a practical standpoint. But don't go overboard that you have to keep it below some artificial limit well below clipping.

[EricBarker]

I worked in the TV industry for 9 stupid years of my life. I had come from the audio world (and promptly returned from whence I came), and thought "Hot = good, as long as it doesn't clip". Let me tell you, the airwave TV industry is S*************TTTTT. I have never seen a bigger load off BS than television. Digital television...self-imposed...a hard limit...of -12dB. Say that out loud. Now go drink until your blood alcohol level matches that level of stupidity. That's over half of 16-bits GONE. TV deals in 8bit audio, because they don't know any better. I guarantee, every sample of TV audio begins with "0000 0000". This is long before congress pasted the LKFS restrictions, it's just "the way it is". This isn't tracking, this isn't practicality, THIS IS MASTERED AUDIO!!!

 

It took many nights at the bar, and a lot of gin for me to go "f*** it" and follow my mentors like a good boy. I spent 9 years, like a Tibetan monk seeking the ultimate truth, trying to figure out what I was missing, some logic to it all, and here's what I came up with:

"TELEVISION IS RUN BY MONKEYS!"

 

Next time you have to turn on CC to be able to hear what Jon Stark is saying... the more you know. Excuse me, I have a handle of rum to finish.

[Reezekeys]

Hypothetically... if you NEVER get a digital over, is there ANY reason to record anywhere below -0.5dB?

If I can hazard a guess, I think recording engineers just don't see much payoff in improved sound quality or s/n by spending extra time obsessing over getting record levels as hot as possible (like they might have done with tape). Setting them so the average meter reading is -12 pretty much guarantees no overs, while still keeping levels high enough off the noise floor to not be an issue.

 

HOWEVER, I've heard people swear rounding errors from hearty attenuation causing distortion. This seems extremely likely in a float environment.

I always thought floating point allowed for more precision, which should minimize rounding errors. Do I have this wrong?

[Ulf]

I worked in the TV industry for 9 stupid years of my life. I had come from the audio world (and promptly returned from whence I came), and thought "Hot = good, as long as it doesn't clip". Let me tell you, the airwave TV industry is S*************TTTTT. I have never seen a bigger load off BS than television. Digital television...self-imposed...a hard limit...of -12dB. Say that out loud. Now go drink until your blood alcohol level matches that level of stupidity. That's over half of 16-bits GONE. TV deals in 8bit audio, because they don't know any better. I guarantee, every sample of TV audio begins with "0000 0000".

 

Care to explain your math here? If I'm not mistaken, the -12dB down from the ideal 96dB (at 16 bits resolution), would mean "wasting" ~2 bits.

 

If you don't believe the math you can easily listen for yourself by recording at -12dB and compare to a 8 bit recording.