Quantcast
Extra sample rate conversions because of plugins ? - Gearspace.com
The No.1 Website for Pro Audio
Extra sample rate conversions because of plugins ?
Old 28th January 2009
  #1
Gear Head
 
🎧 10 years
Extra sample rate conversions because of plugins ?

Hello there, I did search a lot in the forum but didn't find the proper answer to my question... ...so I hope this hasn't been covered yet...also sorry for my bad english ...writing from Switzerland....

I record on tape and tranfer to 24bits/96kz for mastering. In the end I have to sample convert to 44.1hz and dither to 16 bits (I understood this is the right order). The problem is that between sample convert and dithering I have to do some level adjustments.

For example using RBrain to sample convert will leave some overs. If you master with lower levels so that there is no overs after RBrain (or choose the option prevent clipping in RBrain) you have a file that can be 1db quieter just because of handfull of overs.
Even if I'm not in the loudness war, it seems stupid to leave it like that and I will try to regain this 1 db by going for example through elephant and by the same time dithering to 16bits.

Here is the problem: Elephant works better in oversampling, which I understand is upsampling the file, processing it and downsampling again.

So I think by this process I downsampled twice my file which I want absolutly to avoid as I believe too many SRCs are responsible for loss of quality in digital.

How can I avoid the second downsampling ? Should I leave Elephant without oversampling or the second downsampling in elephant not so critical for a reason that I don't understand....? Is there another better process to go....?

Thanks in advance....

Juan
Old 29th January 2009
  #2
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Interesting. I go from tape to 88.2KHz 24-bit. When I'm done mastering, I'll normalize to -.2dB, downsample, dither & truncate. I never get any overages. Perhaps it's an error in your sample rate convertor. Or perhaps since you're using an assymetrical downconversion (what's the least common multiple of 44.1 and 96?), that may be creating some stray peaks.
Old 29th January 2009 | Show parent
  #3
Gear Addict
 
Nishmaster's Avatar
 
🎧 15 years
If any of the program transients have been clipped or totally brickwalled, resampling may restore portions of those peaks, thus causing your overages.

Since you had clipped those peaks before, it's not going to hurt much further to do it again. Don't worry about leaving headroom for overages after resampling. Just take your file after resampling and normalize it to -.2dBFS, and you'll be fine.
Old 29th January 2009 | Show parent
  #4
Mastering
 
🎧 15 years
Quote:
Originally Posted by Nishmaster ➑️
If any of the program transients have been clipped or totally brickwalled, resampling may restore portions of those peaks, thus causing your overages.

Since you had clipped those peaks before, it's not going to hurt much further to do it again. Don't worry about leaving headroom for overages after resampling. Just take your file after resampling and normalize it to -.2dBFS, and you'll be fine.
I disagree. Depends on your definition of "hurt". It's probably going to be subtle, but the fact is that when changing domains, the signal level can increase (as little as 0.2 dB, as much as 3 dB in extreme cases) and this can cause clipping and this affects the sound, reduces the impact and the dynamics and adds distortion. It's a fact of life. The Weiss Saracon and SFC-2 include a default level drop within the algorithm. You have to close the barn door before the horse escapes, it doesn't help to do it after!

Crossing domains, examples:

---- conversion to mp3. Drop the level prior to the conversion if the source material contains lots of high frequency peaks and/or full scale peaks as the mp3 is likely to overload. This is due to the filtering in the mp3 encoding causing over levels.

---- conversion to analog. The filtering in the converter can produce what are sometimes called "0 dBFS+" signals. Solution: reduce the level prior to conversion.

--- sample rate conversion. similar...

Is it subtle? Yes and sometimes no. Is it audible? Yes and sometimes no. Is it a problem? Yes, if you ignore it. Use an oversampling peak meter to measure the true peak levels.


BK
Old 29th January 2009 | Show parent
  #5
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
That reminds me, I have a piece of software that can attempt to restore clipped audio. As you might expect, it seems to do more damage than good but it IS about 7 years old. Do you know of any software out there that can fill in the missing information? Sometimes I get clipped mixes and can't get a remix done. In order to hide the distortion, I have to use EQ, add soft clipping (to round off the edges of the plateaus) or just do nothing. I know you can't restore it to its original quality ever but I was just wondering if any advancements have been made in this area.

Bob K, would you suggest doing the SRC in stages? Like downsample to 44.1KHz, do any final level adjustments, then dither to 16-bits?
Old 29th January 2009 | Show parent
  #6
Mastering
 
🎧 15 years
Quote:
Originally Posted by wado1942 ➑️
That reminds me, I have a piece of software that can attempt to restore clipped audio. As you might expect, it seems to do more damage than good but it IS about 7 years old. Do you know of any software out there that can fill in the missing information? Sometimes I get clipped mixes and can't get a remix done. In order to hide the distortion, I have to use EQ, add soft clipping (to round off the edges of the plateaus) or just do nothing. I know you can't restore it to its original quality ever but I was just wondering if any advancements have been made in this area.

Bob K, would you suggest doing the SRC in stages? Like downsample to 44.1KHz, do any final level adjustments, then dither to 16-bits?

If the SRC outputs in floating point and you save to a floating point file, you can leave the level drop till after the SRC. But if not, then you can't get away with skipping the requirement to drop the level BEFORE the SRC.

There are some declippers. They all do some form of interpolation. I've heard the Cedar declipper on headphones at an AES convention and it was impressive, but this requires the Cambridge hardware platform, it's not a standalone (damn!). Reportedly the Cube Tec system has some powerful standalone VST plugins now, one of which does declipping, but I have not auditioned it. Izotope has a declipper but I have not auditioned. Sonic Solutions' E type and B type does a nice job with short duration clips. Some people report success using a decrackler as a declipper but am not impressed. Also, remember to leave headroom for the restored audio levels. Good luck, your mileage may vary!

Bottom line: I don't own a good-sounding declipper yet... :-) But I'd love to see one in VST that sounds good and works well.

BK
Old 29th January 2009 | Show parent
  #7
Lives for gear
 
Tube World's Avatar
 
5 Reviews written
🎧 10 years
How is the declipper in Sequoia?
Old 29th January 2009 | Show parent
  #8
Lives for gear
 
wado1942's Avatar
 
🎧 10 years
Yeah, my software is all 32-bit or 64-bit float. What I've been doing is running through a limiter at 88.2KHz (set to a max -0.2dB), then downsampling to 44.1K 16-bit and haven't had any overages yet. If I DID have overages, I could just undo it and reduce the level by whatever amount.

You know, it shouldn't be too hard to DESIGN a half-way usable declipper. It wouldn't be much different than an oversampling converter. The only challenge would be identifying where the clipped audio is and using valid samples to the side of the clipped portion and continue those trends over the clipped portion.

But you're preaching to the choir about headroom. I keep everything around -20dBfs or so till right before downsampling to 16-bit (for CD). I get E-mails all the time from people asking for advice on better mixes etc. Their problems are almost always from (a), improperly placed and calibrated monitors or (b) running their levels too hot. Generally speaking, if their running their levels too hot, it's because their monitors aren't calibrated... Or they think in order to have a loud CD, everything has to be tracked super-hot, mixed super hot and then made even hotter on mastering. Which, as you know, just makes for a distorted mess that CAN'T be made loud.
Old 29th January 2009
  #9
Gear Head
 
🎧 10 years
The fact that RBrain leaves some overs seems to be normal as this is quoted in their website:

"...the sample rate conversion process often adjusts peak structure of the original program material, thus, in many cases, making a subsequent peak-limiting a necessity. "

My question is not so much about the clipping, as I don't normally get any or very few, but about "...making a subsequent peak-limiting...". If I do this with a limiter that has oversambling, I think I'm basically upsampling my 44,1 kh file (after Rbrain) and downsamplig back to 44,1 kh after the limiting....

I want to avoid downsampling two times to 44,1 hz.
I understand this is a critical process from the many threads about "Which is the best downsampler"...So what if you use the best downsampler but then put your file through a limiter that will upsample and downsample again...? What about the quality of this internal downsampling in the limiting plugin ? Is this for some reason not so critical as the first downsampling of the file ? Or would it be better to use the limiter without the over sampling mode....?
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 56 views: 17319
Avatar for Alexey Lukin
Alexey Lukin 3rd March 2021
replies: 57 views: 15711
Avatar for Alexey Lukin
Alexey Lukin 8th December 2011
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump