dither problem

  • Thread starter Thread starter Rusty K
  • Start date Start date
R

Rusty K

New member
Hello,

I'm getting over a 2db increase in amplitude on a particular master when I convert down to 44100/16. Before this particular track the most I've ever increased in aplitude is about 1db. Does anyone have a clue as to why the dither process is causing this much of a increase?

I had read in another thread that by keeping the master to about -1db this should be enough headroom to allow for dithering without causing a clip....not so in my case.

Thanks,
Rusty K

PS I might add that the pre master mix had a couple spikes over -1db.
 
I don't get any change in amplitude at all, ever, when I dither down. I don't think you should be getting any change either. The only thing that should change is the bitdepth and the sample rate, right?

However, I record and mix at 44,100 and 24-bit, so when I dither down, it's only the bitdepth that changes. Yet I don't think changing the sample rate should alter amplitude.
 
I have read on another thread that you should leave at least 1db headroom because dither will add amplitude (usually less than 1db), I'm sometimes getting over 2db.

I've had the "save non audio information" box in "Save As" checked for no reason. Could this make some kind of difference? I'm grasping at straws here. I'll keep working with my software to see if I can figure out what it is I'm doing to cause this.

I am "squeezing" the signal a bit to get all the amplitude I can and still have a realtively transparent sound but that's pretty normal isn't it?

It only seems to be a problem on some waves, some of those generated on my PC. Could it be plugin related?

Some of my pre master mixdowns may be a tad hotter than -1db say in one channel. Could that be the cause? Should I keep all spikes in a pre master mixdown below -1db?

Thanks,
Rusty K
 
dobro said:
I don't get any change in amplitude at all, ever, when I dither down. I don't think you should be getting any change either. The only thing that should change is the bitdepth and the sample rate, right?

However, I record and mix at 44,100 and 24-bit, so when I dither down, it's only the bitdepth that changes. Yet I don't think changing the sample rate should alter amplitude.
Click to expand...
Once again, dobro, I find myself in the same boat as you. However, I record at 32bit-float at 44100... however, I have my dithering disabled.

I don't dither..... I think :D

I simply record, and save the session. I save the files as PCM so they maintain their respective sample rate as which they were recorded (which can only be 24-bit due to the soundcard)

Upon final mix, I export the audio file as a PCM wave.... So here's the only real question, am I actually maintaining my 24-bit depth to my recorded vocals when I save them as PCM wave? What about exporting the final mix?

If it does maintain it's recorded bit depth, then I suppose there is no dither at all in my process... right?

I do know, that in my properties window, or whatever, that dithering is disabled.
 
My understanding is that until you go into Edit>Convert Sample Type, you're continuing to operate at 32-bit floating point without any dithering.
 
dobro said:
My understanding is that until you go into Edit>Convert Sample Type, you're continuing to operate at 32-bit floating point without any dithering.
Click to expand...
Okay, but Exporting to wav can't maintain 32 bit can it? 24 bit I can see, but it seems likely it would be 16 bit Audio...

As a matter of fact, when I click on a finished master file, Adobe recognizes it as 16bit audio @ 44100 hz...

So, exporting to wav does convert to 16 bit...

Therefore, what's the purpose of recording in 24 bit, if the wav format doesn't allow that bit depth?
 
I don't know about Exporting to Wav, cuz I've never done it. I only know about tracking and editing, which is all best done in 32-bit floating point, I think. Then dither down as the very last step when you're mixing down.
 
We use .wav to make audio CD's, which are 16bit 44hz.

You track and mixing at 24bit or "32bit float" in CEP to reduce quantization error from volume, eq and effect changes.

As soon as you change anything to your track in CEP, it multiplies and rounds the bit value to the nearest whatever. If you record in 16bit, you will hear the rounding error as distortion during mixing....not so much the case with 32-bit.

I'm sure that CEP adds dither automatically when you mixdown a file to .wav.

So mix in 32, then mixdown to .wav, master and burn. If you want to master at 32, which would make a lot of sense if you believe what I just said above, you'll have to go to CEP Options > Settings > Multitrack and switch "mixdown" to 32 bit from 16 bit.

hope this helps.
 
Rusty, I don't know. My only suggestion is trying a different noise shaping option...I use triangular. What do you use?
 
Rusty K said:
I've had the "save non audio information" box in "Save As" checked for no reason. Could this make some kind of difference?
Click to expand...
I don't think so...I think that's just the little pk files. Not sure though.
 
Chris,

Thanks for getting back to my question. I figured my thread had been "hijacked"...ha.

I'm using triangular/44,100 noise shaping. I've tried a couple of different shapes but it doesn't seem to make much difference.

One thing since I've been snooping around looking for the problem is that I found that "wave analyze" which I have come to rely on all the time is not as accurate about clips as the "gather statistics" function in the clip restoration tool. "Wave analyze" gives "possible" clips. I was assuming this meant I was clipping. I've double checked some tracks with the clip restore plug only to find that the "possibles" were not actual clips.

Still it kind of defeats the purpose of "leveling/squeezing" the wave to gain max amplitude if one has to lower the overall level to allow for a 2db spike when dithering.

I'm still looking...can you think of a forum or technical member here who might have the answer? I have posted this thread in "Mixing and Mastering" but I haven't gotten the answer yet.

Thanks,
Rusty K
 
Hey...maybe "sonusman" over at www.recordingproject.com. He knows everything, lol.

You know, when I do that dither down to 16bit, I see the waveform change...but I've never even paid attention as to whether it's getting louder in places...if I had noticed it, then I would probably do one extra step after dithering (a no no) and limited the mix to -.1db with no output gain at all. The biggest question is, can you HEAR it clipping after dithering? Keep in mind, CEP's meters are set a bit hotter than other software, and also keep in mind that hitting the red doesn't necessarily mean you clipped...it only means you hit 0db...(or really, since it's cep, you hit about -.25db).
 
chrisharis,

Well your last reply to my thread just placed you in contention for one of the most informative.

I've been wondering about sonusman..hadn't seen him around. He's responsible for me purchasing my LynxOne 24bit soundcard, love it! Does he mind someone showing up and asking a question?

Thank for the heads-up on how CEP functions, I didn't know any of that.

Rusty K
 
Rusty K said:
I've been wondering about sonusman..hadn't seen him around. He's responsible for me purchasing my LynxOne 24bit soundcard, love it! Does he mind someone showing up and asking a question?
Click to expand...
He absolutely would not mind. He mostly hangs out over at RP these days. If you haven't done so, sign up and just pm him. He uses the same name over there, and he just LOVES to explain all the bit/byte crap that I never have paid enough attention to, lol.

Best of luck.
 
You must log in or register to reply here.
Back
Top