Is this statement true?

[billisa]

Obviously, digital isn't like being able to pin the meters in analogue, but I'd heard/read that you should still track as close to 0 or -1 as possible, without exceeding, or you're not using all 16 bits... Then I came across this comment...

"In the digital realm you need to track at -12DB and mixdown just a little above that.Mastering houses like to have a lot of headroom to work with,so a final (stereo 2 track) mix at -12 DB wouldnt be a bad idea either.You're still using all the available bits. Anything above 0 db in the digital world is called digital distortion/Clipping."

Is the -12db part true?

 

[chessrock]

I think as long as you don't go over . . . and as long as you don't use any processing on the master bus; like EQ, compression or limiting . . . then any mastering engineer should be able to work with it.

 

[eeldip]

noise with digital is just silly low. so dont feel back tracking at really low levels and raising the level later. no biggie.

your average speaker has a dynamic range of 85db. your average computer system has WAY more dynamic range than that. lots of room to fuck up.

 

[TexRoadkill]

noise with digital is just silly low. so dont feel back tracking at really low levels and raising the level later. no biggie.

your average speaker has a dynamic range of 85db. your average computer system has WAY more dynamic range than that. lots of room to fuck up.

Not really. Most cheap digital stuff has an SNR in the 80's to low 90's. I would still track pretty hot on cheap 16bit gear. If you have a nice 24bit system with good convertors then it's not so crucial to track hot.

 

[mshilarious]

-12dB seems a little conservative. It ain't that hard to shoot for -3dB if you're worried about clipping.

Since I only do my own stuff, I can generally track hot as I know I'm not going to peak higher than I do when I set levels. That's a bit tougher with inexperienced musicians; they don't give you a real hot sound check. If I had to deal with that a lot, I'd get an outboard compressor.

 

[Farview]

using all the bits

Hear is the story the way I remember it. 1 bit is equal to 6db, so as long as your mix peaks above -6 dbfs, you have used all 16 bits. The story about mixing at -12dbfs came from the good old days when we were using analog meters. 0db on the meters on your mixer is line level and should be equal to -12dbfs on your digital equipment. so, when you mix your masters, you need to average around -12dbfs. Just like when you were setting the levels on your cassette deck, you set it so the meters bounced around the 0 db mark (0dbu = -12dbfs)
jason
www.farviewrecording.com

 

[billisa]

Well, by and large, what I've been doing is tracking to within a few db of 0, then mixing down with heavy transient peaks sometimes reaching 0, but never exceeding... I know clipping when I hear it, so I can easily avoid it.

I think it was Frank Filipetti who said you should track as close to 0 without exceeding, because if not, then you're not using all 16bits. Then I ran across the -12db suggestion...

On mixdown, my Fostex VF160 seems to have a safety zone built-in because I've had times when the meter bumps past 0 without any discernable clipping. But still, I just keep it a tad under...

 

[HangDawg]

I guess the best thing to do is switch to 24bit gear and not worry about it.

 

[Ethan Winer]

Dawg,

> I guess the best thing to do is switch to 24bit gear and not worry about it. <

Even with 16-bit gear you don't necessarily have to worry about it. I did a test of exactly this issue a while ago, and anyone here who's interested can download and listen to the same acoustic guitar track at various bit depths. Here's a direct link:

www.ethanwiner.com/BitsTest.html

The files are 16 bits, 13 bits, 11 bits, and 9 bits. You can email me with your guesses. And yes, I do mean guesses.

--Ethan

 

[fivestar]

Dawg,

> I guess the best thing to do is switch to 24bit gear and not worry about it. <

Even with 16-bit gear you don't necessarily have to worry about it. I did a test of exactly this issue a while ago, and anyone here who's interested can download and listen to the same acoustic guitar track at various bit depths. Here's a direct link:

www.ethanwiner.com/BitsTest.html

The files are 16 bits, 13 bits, 11 bits, and 9 bits. You can email me with your guesses. And yes, I do mean guesses.

--Ethan

i tried that test the other day. i could barely hear any difference in the ones i could tell were lower ones (maybe 2 of them) and the rest, i couldn't tell a difference in at all. definately not worth the extra hard drive space, IMO.

 

[Farview]

The only thing more bits allows you to do is get quieter. The added bits just push the noise floor down further.

 

[dirtythermos]

The only thing more bits allows you to do is get quieter. The added bits just push the noise floor down further.

Because of the way a computer does division, extra bits allow you to effects processing with fewer artifacts.

 

[t glyptic]

As I understand it -12 in the digital world is 0 in the analogue world. Digital systems can't go higher than 0db (to spite the games your meters play on you). I could be way off base, that's just what some else with a lot of experience shared with me.

I always think it's good to track at about -12rms just so I have lots of headroom. I can do more with it. If you're sending your stuff ot a mastering house you probobly should leave the "squishing" up to them.

I've heard some good ideas on this post. Intersting stuff.

 

[Farview]

Because of the way a computer does division, extra bits allow you to effects processing with fewer artifacts.

That is why you need more bits internally, for processing. Most of your processing on computers is done at at least 32bit no matter what the original file was recorded at, or how many bits the end result will be.

 

[TexRoadkill]

Dawg,

> I guess the best thing to do is switch to 24bit gear and not worry about it. <

Even with 16-bit gear you don't necessarily have to worry about it. I did a test of exactly this issue a while ago, and anyone here who's interested can download and listen to the same acoustic guitar track at various bit depths. Here's a direct link:

www.ethanwiner.com/BitsTest.html

The files are 16 bits, 13 bits, 11 bits, and 9 bits. You can email me with your guesses. And yes, I do mean guesses.

--Ethan

If you used good 24bit convertors in 16bit mode then the SNR is probably still pretty good. I would assume a bigger difference using cheaper convertors that max out at 16bits.

 

[Ethan Winer]

Folks,

> The only thing more bits allows you to do is get quieter. The added bits just push the noise floor down further. <

Yes, though more bits can also lower distortion. But distortion is very low in a 16-bit system, even at 12 dB below 0. It's far lower than even the finest analog tape recorders.

> Because of the way a computer does division, extra bits allow you to effects processing with fewer artifacts. <

Yes, but the artifacts you refer to are simply the same distortion I mentioned above. It's true that every digital operation, even a simple volume change, increases distortion. But the increase is very tiny. I urge you to experiment with this for yourself. Take a great sounding Wave file and bring it into SoundForge or another basic audio editor program. Lower the volume by 2 dB, save the file, then raise the volume by 2 dB, save again, and repeat that 10 or even 50 times. Listen to the file every few passes and see if/when you can hear the quality degrade.

> Most of your processing on computers is done at at least 32bit no matter what the original file was recorded at <

Yes, excellent point. And this is exactly why using 16 bits is not nearly the compromise that many people think it is.

What amazes me about these discussions is seeing people argue about 0.01% distortion versus 0.001% when they're listening in an untreated room where the frequency response varies by 30 or more dB throughout the entire low end. Priorities, people!

--Ethan

 

[Farview]

What amazes me about these discussions is seeing people argue about 0.01% distortion versus 0.001% when they're listening in an untreated room where the frequency response varies by 30 or more dB throughout the entire low end. Priorities, people!

Exellent point! There is a whole population of people that think mp3s sound good, and we are wasting time picking the smallest of nits.

 

[chessrock]

Well, ya' know . . . most 1/8" tape machines sound just fine to me, and do a more than respectable job, too. That doesn't mean I'm planning on going back to the Tascam 4-track cassette portastudio any time soon.

Multiply these miniscule artifacts over a 10-20 track mix, sum it, and it starts to become more apparent.

 

[NL5]

There is a whole population of people that think mp3s sound good, and we are wasting time picking the smallest of nits.

Good point. There is nothing more frustrating to me than getting just "that sound" you were going for, then A/B'ing it for the wife, only to hear - "they sound the same to me"....arghhhhhh.

 

[Farview]

Well, ya' know . . . most 1/8" tape machines sound just fine to me, and do a more than respectable job, too. That doesn't mean I'm planning on going back to the Tascam 4-track cassette portastudio any time soon.

Multiply these miniscule artifacts over a 10-20 track mix, sum it, and it starts to become more apparent.

Those artifacts are still much more subtle than the hiss of a 16 track 1/2 inch machine, or even a 24 track 2 inch without dbx (because it does funny things to the transients) Between the hiss, distortion, and softening of the transients (which will keep getting softer as time goes by) it's a wonder we could hear the music.

The only reason we can hear the errors in digital recordings is because the recording is clear enough to allow you to.