attenuation of high frequencies when mixdown

  • Thread starter Thread starter SuperBlade
  • Start date Start date
Hmm. I did the test and did notice the attenuation. It is easily solved by making sure dithering and the pre/post filter are enabled. But what I'm confused about now is *why* that is even neccessary when you're in 44.1/16 bit the whole time.

I'm obviously missing something here.

Chris
 
Hey Groucho...can you tell me what settings you used for the dithering...I seriously put mine on 999, and I'm thinking that might be overkill.

And what the hell is a pre/post filter...lol. God, I'm worse off than I thought.

thanks,
chris
 
Chris:

If you haven't started experimenting with dithering & noise shaping, my advice is to download the Ozone dithering guide and spend a few days figuring it out. I had never paid much attention to it until recently, but it makes a significant difference when you convert your final mix to 16 bit (assuming you've been working in 32-bit float up until that point, which you should be).

I only work in 44.1 and I always mixdown to 32 bit, so I've never had any reason for dithering during the mixdown (& since I stay in the same sample rate, the low/high slider isn't a factor for me). In general, it's a good idea to keep your sample rate and bit depth as high as possible until the very end, when you convert to 16-bit for CD burning purposes.

I usually leave the dither depth at 1 point & use the triangular pdf shape. But the noise shaping profile to use really depends on the material you're working with. Try different ones and see which sounds best.

What noise shaping does, basically (apologies if you know this) is shift the added frequency noise that occurs during dithering to areas of the spectrum that are less likely to be noticed. I tend to use the C2 shape, because it shifts the noise away from the upper mids area the most and keeps the high end smooth and dandy the way it sounds in 32 bit before the conversion.

As far as the pre/post filter: again, this is something that's only neccessary if you're changing sample rates, but it keeps false frequencies from being generated at the low end. It should always be enabled.

But what I don't understand is why, when "mixing down" a 44.1/16 bit file to another 44.1/16 bit file, there should be any change whatsoever. I posted this question to the CE forum. We'll see what they say.

Chris
 
One thought - when you load a file into CEP, doesn't it get converted to 32 bit, no matter what the source file? If so, you would probably get good high frequencies if you 1) truncated down to 16-bit as long as every point in the chain was set to no changes at all (volume, trim, etc -- all at exactly zero dB), or 2) applied good dither. The middle ground of quick and dirty dither could hurt your high frequencies.

-lee-
 
No, you have to specify that in the options (if you want all files to be automatically converted to 32 bit). The situation at hand is only involving 16 bit files.

I'm still waiting on a response over at the Syntrillium forum. Getting curious about this one.

Chris
 
I actually care about this topic, so I spent some time pouring through that thread.


:confused:


I'm sure everything those guys said had some merit and value, but I have no idea what 99% of it means. All I wanna' know is, how do I not kill the highs, right? I was going to post in that forum to address the comments about "not hearing" the difference, but I think it would have been a waste of time. It's obvious to me that I've got some setting enabled that shouldn't be, but all the discussions about

One thing I am absolutely certain about, though...is that if you take a 16bit file and hit "mixdown" with no other variables, then it changes it...

HERE'S WHAT I THINK - I understand that when you do this, what CEP is doing is actually converting the 16bit file to a 32bit file for internal processing, then BACK to a 16bit file (this is where the dithering happens). So theoretically, if this is right, then the only way to avoid it would be to record in 32 bit, do premixes in 32 bit, and mixdown to 32 bit. That's the only way CEP won't be dithering and re-dithering (whatever the fuck that is) numerous times...it'll do it once, on the final mixdown to 16bit.

I'm frequently wrong about these things, though. Groucho, you started that thread, and you at least typed near the end like you understood what was going on. AM I RIGHT???

For the record, I typically DON'T have all the 32bit boxes checked, b/c of hard drive space limitations. But I frequently have as many as 6 drum tracks open that I've gotten via CD (16bit), so I figure why convert it to a higher bitrate?? Well, the reason would seem to be because CEP does it anyway (transparently) and then BACK to 16 bit, so there's no telling how much "dithering" noise I've been introducing into the process.

FROM THE SYNT FORUM: Basically, it doesn't matter how you go about doing this, whether you use multitrack or edit view, if you have two instances of the same information in two separate files, they will always differ by the uncertainty of the last bit, because of the fact that regardless of the supposed integer format of the files, internally they are still both converted to a FP format. The noise you get by doing this via the multitrack view is no different from the noise you get by adding them in edit view. The noise is inherent in the file saving process. If you didn't dither, then you'd end up with a file that may ultimately rsolve to a flat line, but which would have all of the problems associated with an undithered 16-bit file.

Stuff like that helps me none. I'm stoooopid. Can anybody tell me if I should have the options set to "open all files in 32 bit" and all the other 32bit options enabled to avoid this?

thanks,
chris
 
Well, (as I just posted over at the Syntrillium forum) I'm satisfied that I understand what's happening now. Yes, CE is always working in an internal 32-bit format, which explains why, when mixing one 16-bit file down to another, enabling dithering eliminated any change in the sound for me.

However, as I've said from the beginning, this technically shouldn't even be an issue since if you're concerned about retaining as much fidelity as possible, you should always be working in 32 bit right up to the end. I can't really imagine a scenario when I would choose to mixdown to 16-bit, since anything I mix down is going to have some kind of transform done to it at some point, in which case I'm going to want it to be in 32 bit.

>But I frequently have as many as 6 drum tracks open that I've >gotten via CD (16bit), so I figure why convert it to a higher >bitrate?? Well, the reason would seem to be because CEP does >it anyway (transparently) and then BACK to 16 bit, so there's no >telling how much "dithering" noise I've been introducing into the >process.

Even if the source file is in 16 bit, you're much better off converting it to 32 bit, because otherwise any transforms you do to the file will cause the sample info to be truncated - and therefore you're losing information. That's the whole point of a "floating point" file. And the fact that CE's internal format is 32-bit (which is a good thing) is another good reason for staying in 32 bit the whole time.

So, to sum up, if you want to mixdown to a 16-bit file, just make sure you're enabling the dithering/noise-shaping option & then all is well. But for best results, just stay in 32 bit and make the conversion to 16-bit (along with dithering and noise-shaping) your absolutely final step, after you've done everything else to the file.

Chris
 
Okay, I know this'll piss somebody off, but I'm really NOT being lazy when I ask this. GROUCHO...please tell me EXACTLY how and where you have your 32 bit things configured. In "settings" or "options" or both? I actually have the CEP 2.0 manual (i.e., not pirated), but I can't seem to get this right.

1. "Record in 32?"
2. "Open waves in 32"
3. "Mixdown to 32"????

All of the above, plus others???

Thanks,
Chris
 
Chris, I can't help with the settings since I dont have 2.0 but just in case it wasn't clear for anyone you want to mixdown to 32 bit then convert that final stereo mixdown file to 16bit for CD Burning.

Be sure to save a copy of the 32bit mixdown so you can make a higher quality DVD Audio version down the road when that becomes cheaper to do.
 
Thanks, Tex. Is it okay for the "premixing" to be in 16bit? I mean, that's just what you're hearing in the multitrack, right?
 
chrisharris said:
Okay, I know this'll piss somebody off, but I'm really NOT being lazy when I ask this. GROUCHO...please tell me EXACTLY how and where you have your 32 bit things configured. In "settings" or "options" or both? I actually have the CEP 2.0 manual (i.e., not pirated), but I can't seem to get this right.

1. "Record in 32?"
2. "Open waves in 32"
3. "Mixdown to 32"????

All of the above, plus others???

Thanks,
Chris
Click to expand...

Well, I don't *record* in CE (I do it on an external hd recorder and dump the tracks into CE later), but, assuming your sound card is capable of 32 (or 24) bit recording, you'd want to have "record in 32" checked.

I don't have "open all waves in 32 bit" checked. I simply open them, then covert them to 32 bit. Somwhere in the manual there's a line that mentions why this is a better way to go - something to do with the manual conversion process being preferable to having it automated upon opening, though I confess I can't remember offhand why.

Yes, I have "mixdown to 32" checked. Pre-mixing is at 32. And make sure "dither transform results" is checked (I believe it is by default).

On the "data" tab I have everything checked except "open all files as 32 bit".

hope that helps,
Chris
 
chrisharris said:
Thanks, Tex. Is it okay for the "premixing" to be in 16bit? I mean, that's just what you're hearing in the multitrack, right?
Click to expand...

I'm not really sure what they mean by 'pre-mixing'. To be on the safe side I would just stick with 32bit for EVERYTHING until you are ready to burn.
 
You must log in or register to reply here.
Back
Top