You are here

Multi sampling. 16 or 24 bit

All about the tools and techniques involved in capturing sound, in the studio or on location.

Re: Multi sampling. 16 or 24 bit

Postby Hugh Robjohns » Sat Feb 01, 2020 5:48 pm

Kolakube wrote:I didnt see where Hugh said 'Triangular' dithering.

Sorry -- TPDF = Triangular Probability Density Function -- the proper name for bog standard 'triangular dithering'. it provides a uniform noise floor with a flat frequency response much like analogue electronics.
User avatar
Hugh Robjohns
Moderator
Posts: 26331
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Multi sampling. 16 or 24 bit

Postby Hugh Robjohns » Sat Feb 01, 2020 6:00 pm

Kolakube wrote:There are two types of Bit Depth Conversion. One is erm, just that (We will call it vanilla) and the other is POW-r So right at the offset im drowning in jargon. Before I even get to the dither page what Bit rate reduction would you recommend I use?

If we're talking Jargon, the correct term is word-length reduction... because you are reducing the word from 24 bits in length to 16 bits... But it's the 'word' as a whole that's getting shorted. So not 'bit depth' or 'bit rate'...

As for POW-r, it's juts a proprietary noise-shaped dither algorithm which effectively scoops noise from the region where our ears are most sensitive, and dumps it in areas where they're not -- up at the high end! So potentially a noise-shaped dither will sound subjectively quieter than a flat TPDF dither, but against that there is an increased risk of the raised high-frequency dither energy interacting with the programme material in an adverse way... To be honest, the differences are extremely subtle in most applications and really only become relevant if you're working with source material that has an extraordinarily large dynamic range.

This plot illustrates the noise spectrum of POW-r compared to standard TPDF (triangular) dither:

Image

...both dither menus then have Noise Shaping. What can then be set too High Pass contour or Equal loudness contour or Off? What do you suggest?

I'd leave it set to Off. It would make no sense to further process POW-r as it's already psychoacoustically optimised.

H
User avatar
Hugh Robjohns
Moderator
Posts: 26331
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Multi sampling. 16 or 24 bit

Postby Hugh Robjohns » Sat Feb 01, 2020 6:10 pm

Kolakube wrote:But does me sampling a fart or a blip from my SH-101 need such dynamic range to be successful.

You said you want to make the best possible 16 bit samples for your sampler.

So, if we want to end up with 16 bits, it would be best to start at 24 bits so that we have some space to process and optimise the recorded sounds.

So, if you record at 24 bits you can leave a good 20dB of headroom if you want, without quiet stuff getting anywhere near the noise floor of the A-D converters, and that means you'll never get caught out with your farts or blips overloading the converter either. You can just hit the record button, and fart or blip away to your heart's content happy in the knowledge that you'll always have a clean recording.

Once everything is recorded and chopped up into individual files, you can use a batch normalise routing to set the peak levels of each clip to your desired level... and then you can use another batch process to reduce the wordlength of each file to 16 bits to fit in your sampler.
User avatar
Hugh Robjohns
Moderator
Posts: 26331
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Multi sampling. 16 or 24 bit

Postby Eddy Deegan » Sat Feb 01, 2020 6:34 pm

Kolakube wrote:I have a very punk attitude to music. I just wana get on with it.

Nothing wrong with that :thumbup:

But I would take the time to accept the advice from Hugh and The Elf even if you don't understand it all (I'm not pretending I do either, though I've learned a lot from them). They are professionals with decades of experience, the effort required to do so in this case is minimal and you'll get a better result.

Win win :D
User avatar
Eddy Deegan
Frequent Poster (Level2)
Posts: 3506
Joined: Wed Sep 01, 2004 12:00 am
Location: Brighton & Hove, UK
Some of my works.
Please consider supporting the SOS Forum Album project.
 

Re: Multi sampling. 16 or 24 bit

Postby The Elf » Sat Feb 01, 2020 8:52 pm

Only just returned from a day of pitch editing - phew! :headbang:

Yes, triangular - TPDF.

OK, I'll admit I sometimes use a more sophisticated shaped dither, but that's just to make me feel clever! In Cubase I just drop in UV22, set it for auto-black and never think about it again...

Really, this is not worth agonising over - just record 24-bit, dither with anything simple down to 16-bit and go have a cuppa!

Here's the water Dobbin - the rest is up to you! ;)
User avatar
The Elf
Jedi Poster
Posts: 13767
Joined: Tue Aug 14, 2001 12:00 am
Location: Sheffield, UK
An Eagle for an Emperor, A Kestrel for a Knave.

Re: Multi sampling. 16 or 24 bit

Postby Kolakube » Sat Feb 01, 2020 10:45 pm

@Hugh


There are two types of Bit Depth Conversion. One is erm, just that (We will call it vanilla) and the other is POW-r So right at the offset im drowning in jargon. Before I even get to the dither page what Bit rate reduction would you recommend I use?


If we're talking Jargon, the correct term is word-length reduction... because you are reducing the word from 24 bits in length to 16 bits... But it's the 'word' as a whole that's getting shorted. So not 'bit depth' or 'bit rate'...

Hey....Tell Soundforge. Its their menu. You have to go Process > Bit Reduction ..... :angel:

Re triangle Dither and Pow-r Thanks, ill just use the standard after your explanation

So, if you record at 24 bits you can leave a good 20dB of headroom if you want, without quiet stuff getting anywhere near the noise floor of the A-D converters, and that means you'll never get caught out with your farts or blips overloading the converter either. You can just hit the record button, and fart or blip away to your heart's content happy in the knowledge that you'll always have a clean recording.

Once everything is recorded and chopped up into individual files, you can use a batch normalise routing to set the peak levels of each clip to your desired level... and then you can use another batch process to reduce the wordlength of each file to 16 bits to fit in your sampler.

Gold!! Thank you.




Thanks to all who have contributed.
User avatar
Kolakube
Frequent Poster
Posts: 1420
Joined: Tue Dec 01, 2009 1:00 am
Location: Geordieland

Previous