dithering scenario

[jimlongo]

Check this out . . .

If you go from say 16 bit to 24 bit the bottom 8 bits will just be zeros, no? Then if you go from that 24 back to 16 is there any need to dither since what is being cut off are only zeros anyway?

That's what I hope. What do you think?

[Inaam]

FWIW, I think if you do any change at all in the 24 bit domain, from eq to rendering fades to even turning the volume down a dB, that you will lose resolution going back out 16 without dither.

Anyone care to clarify or expand on that? I'm sure if you do a search on dither you'll find lots more detailed info.

Hope this helps

Inaam

[AE]

Originally posted by Inaam:
FWIW, I think if you do any change at all in the 24 bit domain, from eq to rendering fades to even turning the volume down a dB, that you will lose resolution going back out 16 without dither.

Anyone care to clarify or expand on that? I'm sure if you do a search on dither you'll find lots more detailed info.

Hope this helps

Inaam

<font size="2" face="Verdana, Arial">This is true.

It sounds like you may be passing a 16 bit signal through a console with 24 bit resolution and then back to another 16 bit destination. If so, as long as you follow the suggestions above the correct technique is to defeat dither on the console outputs. This is one of those rare situations in which you'll reduce bit length without dithering.

[jimlongo]

I'm just talking about straight digital transfers.

[Barnaby Bristol]

Jim. Correct. If you load a16-bit source into a 24-bit file and then output from the 24-bit file without doing any processing, it should be 16-bit clean. It helps to have a bit meter to assure that you've done it correctly. Things to remember: In Pro Tools, a stereo track is 16-bit clean if the fader is at '0' and both pans are hard left/right. A mono track is 16-bit clean with the pan either hard left or hard right and the fader at '0'. If any other tracks or auxes are assigned to the same output and have instances of plugins that are 48-bit dithered to 24-bit, a bit meter will show 24-bit. You should still be able to truncate without dither since the 24-bit dither is well below the 16-bit level. If you are using the 24-bit optimized mixer "HiRes Dithered Mixer" you will still see 24-bit on your bit meter. If you only have one summing point between your source and output than you should be fine outputting and truncating without dither.

[awestaa]

Barnaby,

What exactly is a 'bit meter' and where do I get one? Is this some specific software product you are using? Where to do you insert it to read bits? Do you use it on files? Well maybe you could just point me to a web site where I can find all this stuff out rather than try and explain....I understand software development and I've used software products on ip traffic streams etc., but I've never heard of this relative to digital audio...it sounds useful.

Thanks,

Andy

___________
www.xen.com

[jimlongo]

Thanks Barnaby. That is helpful.

As far as testing I am asking this question in regards to laybacks which are out of my control. The way it has been explained to me a 16 bit da88 will be transferred to a 20 bit HDCam which will have 16bit DigiBetaCam dubs.

So I am assuming that 16 to 20 to 16 has no truncation distortion during this process.

Thanks again.