question about bit rates

DamagedTrax

just received the multi's for a mix that im doing for someone. they're 16 bit wavs. would it be any benefit to batch process them to 24bit? or should i just leave them as they are and have the final mix recorded to 24bit?

PaulBrewer

DamagedTrax wrote: »

just received the multi's for a mix that im doing for someone. they're 16 bit wavs. would it be any benefit to batch process them to 24bit? or should i just leave them as they are and have the final mix recorded to 24bit?

I've never done that but my gut says stay 16. I've never heard a digital conversion process that improved anything...

DamagedTrax

upwards dithering? never heard of it myself, wonder if it exists? and what would it do?

these are the things i think about when not thinking about the other things that i think about.

PaulBrewer

DamagedTrax wrote: »

upwards dithering? never heard of it myself, wonder if it exists? and what would it do?

these are the things i think about when not thinking about the other things that i think about.

Focus on the other things ....

Neurojazz

In wavelab there is the crystal resampler.... in theory you'd upsample for the benefit of the plugins to run at either their natural 24bit or 32bit algorthims... i.e. lets say you want to stick reverb on something - why feed it 16 bit data and force the clock down? - maybe do a test on a closed hi hat on both 48k24bit and 44k16bit reverbs and see if the 44k is like sandpaper...

At least if that works you can then dither afterwards with that extra quality.

Really depends on your plugins.

PaulBrewer

Neurojazz wrote: »

In wavelab there is the crystal resampler.... in theory you'd upsample for the benefit of the plugins to run at either their natural 24bit or 32bit algorthims... i.e. lets say you want to stick reverb on something - why feed it 16 bit data and force the clock down? - maybe do a test on a closed hi hat on both 48k24bit and 44k16bit reverbs and see if the 44k is like sandpaper...

At least if that works you can then dither afterwards with that extra quality.

Really depends on your plugins.

Changing your raw material for the benefit of extra plugin bits is a very odd way of thinking to my mind.

My experience of converting from anything to anything is it reduces the overall quality, however slightly. The initial bit depth/frequency is what determines what information is captured. Whatever losses are made then are made for good.

Neurojazz

Yeah, i can see that too many conversions are probably a bad thing, but was thinking back to a video i watched where the engineer put reverb over a master to lift it out a bit because it was flat... if the plugin dropped it's processing rate in line with audio it was getting (which it may not anyway) then you'd probably want to upsample (as that quality of reverb would be important to the whole mix)

The rate is probably un-important if using outboard maybe... i'm not sure what the devices would do if connected digitally etc...

I personally wouldn't want to work with 44k, but if i had too would up sample to work with ITB - next time i do, i'll test the reverb in the UAD myself (the wizoo one) - also i'll look that the source wave to see what the difference was in conversion.

PaulBrewer

Neurojazz wrote: »

also i'll look that the source wave to see what the difference was in conversion

SURELY you mean listen?

Neurojazz

Nope - i'll look first, it may seem weird, but i'd probably see the difference close up when comparing the actual wave to see there are any weird conversion bugs.

I often zoom in on the waves to check peaks - should work the same way with testing this theory.... might even just pump a sine wave through and convert that - then the difference would stand out a mile.

PaulBrewer

Neurojazz wrote: »

Nope - i'll look first, it may seem weird, but i'd probably see the difference close up when comparing the actual wave to see there are any weird conversion bugs.

I often zoom in on the waves to check peaks - should work the same way with testing this theory.... might even just pump a sine wave through and convert that - then the difference would stand out a mile.

You and I live on different planets .....

jtsuited

isn't upsampling quite regularly practiced with mastering? granted that's with the sample rate and not bit rate.

Neurojazz

is there a problem with doubling bit rates?

iquinn

The only real benefit I can see would be more headroom when mixing, but there's no real sonic benefit.
Any processing you do ITB, eq/dynamics/gain etc will be working at high floating point resolutions, 32bit/48bit etc, so you're not going to degrade the quality if you stay at 16bit. The internal resolution will preserve the quality that's already there.

With regard to upsampling, often the negatives outweigh any positives and some pluggins are already upsampling to get around problems associated with programming analog processes in the digital domain.

The only good reason to upscale the bit depth or sample rate would be if you're using analog equipment in the mix and doing DA/AD conversions.

In my experience, upscaling vs the original depths/rates in ITB situations, can result in subtly different sounding mixes. But not one better than the other and you'd hear much more of a difference if you tilt your head 1" when listening.

Adding dither when converting a 16bit file to 24bit isn't necessary and would probably make things sound worse. Dither essentially being noise to mask the effects of truncation when going down in bit depth. If you're going up, the samples aren't being truncated, so you're just adding noise that's not doing anything.

godfrey

PaulBrewer wrote: »

You and I live on different planets .....

I blame the magazines for this kind of thing...

PaulBrewer

godfrey wrote: »

I blame the magazines for this kind of thing...

What? Dual Planet Residence?

madtheory

There's no benefit to changing the source file from 16 to 24 bit. The DAW will process the input at whatever the audio engine works at. Most commonly, 32 bit floating point. So the plugins will still process a 16 bit file with 32 bit FP precision.

And in this case, it's bit DEPTH. There's no time element, it's a voltage. High to low and back; hence- DEPTH! Bit rate is used to refer to data streams, which is stuff that takes time to move from one place to another.

Neurojazz

I don't read the magazines.. there's no seller on this planet!

PaulBrewer

madtheory wrote: »

And in this case, it's bit DEPTH. There's no time element, it's a voltage. High to low and back; hence- DEPTH! Bit rate is used to refer to data streams, which is stuff that takes time to move from one place to another.

Yep, it's important that when we are talking about stuff we use the correct terms.

Neurojazz

Yep, understood....

Ok, so the UAD 32 bits plugin depths operation messes with the powercores 24bit operation... UAD even recommend reducing the output/chain level by 3db when your going into the Powercore (as it's 24bit the 32bit signal is too hot)

That make more sense now...?

PaulBrewer

Neurojazz wrote: »

(as it's 24bit the 32bit signal is too hot)

That make more sense now...?

Er, no !

Neurojazz

Ok, if you have a powercore plugin chained after a UAD plugin on your master buss the powercore input level needs to reduced by 3db - i never 100% understood this until recently and can now see why (because of that bit depth difference) - it's probably only an ITB problem that UAD are going on about.

iquinn

Neurojazz wrote: »

Yep, understood....

Ok, so the UAD 32 bits plugin depths operation messes with the powercores 24bit operation... UAD even recommend reducing the output/chain level by 3db when your going into the Powercore (as it's 24bit the 32bit signal is too hot)

That make more sense now...?

Is the powercore not capable of running at a higher bit depth than 24? The sonnox pluggins are 48bit and they can run on powercore.

Neurojazz

i think it's the MD3 that's the issue (the main mastering plugin on there...) - i think it's not been brought up to date for a fair while.... - i think it internally does 24 --> 48 but outputs at 24...

The info came from here - this guy does the mastering tuts for UAD and show the issues with the powercore on the videos (not slating it, just showing being used in the chain [the md3])

iquinn

right o.

I've heard of him before.
I like that http://www.dynamicrange.de/ initiative he's involved with.

Neurojazz

The videos are awesome... he's impartial, but your'e pretty much boned if you don't have the same sort of setup... I was very, very lucky that i'd got most of it already intuitively.

All the lessons are in bite size sections and cover everything upto red book productions etc...

I must sit down again and absorb more - obviously slipping on the details yet again

fabbydabby

madtheory wrote: »

And in this case, it's bit DEPTH. There's no time element, it's a voltage. High to low and back; hence- DEPTH! Bit rate is used to refer to data streams, which is stuff that takes time to move from one place to another.

If that's the case, then why do people refer to the bit rate of MP3s, as in 160kb/s or whatever?

ogy

thats cos that is the bit rate the files encoded at, not the bit-depth, the bit-depth will generally still be 16-bit like CD quality

PaulBrewer

fabbydabby wrote: »

If that's the case, then why do people refer to the bit rate of MP3s, as in 160kb/s or whatever?

The hint is 160kb/s .... per second. Exactly as madtheory described.