Using VSL VI in 96k SR project - good or bad?

[Mystr Tiger]

My questions seems pretty basic - I tried a search but got no hits. So, if this was already covered please excuse and point me to a thread or two. Ok, here's my question:

The (some?/all?) VSL VI samples are 24 bit/44.1k, true? Using them in a 24bit/96k project destined for 16bit/44.k CD makes we wonder whether the conversion from 24/44.1k -> 24/96k -> 16/44.1k will be bad. What does everyone else do? Stay at 44.1k? Or, 96k is no problem? I would think that if the project is 100% VSL with no plug-in processing, then 44.1k would make the most sense. But, if VSL is being used along with other recorded instruments and vocals, and with lot's of processing, well, maybe 96k could be better. BTW, I do not (yet) have a strong opinion one way or the other about 44.1k vs 96k in general - I'm still experimenting - my question is really about the back and forth conversions, plus I'm curious what everyone else is doing since I'm just getting started with VSL VI.

[DM33]

I do all my project at 24 bit / 48khz.

You get more headroom at the higher bit rate and a wider freq response with higher sample rate, and a better transient response. You will also get a lower latency with the a higher sampler rate, around 1-2ms less.

If you use a good dithering plugin or hardware setting your projects to higher bit/sample rates should yield audible benefits but the resultingn audio files will be bigger and you will be placing a heavier load on CPU, harddrives, and ram.

You'll have to see if the extra load is worth it for you.

David

[Mystr Tiger]

Thanks David. I guess 48k might be a nice option too for the audible improvemnet but with only a small increase in resources (as opposed to the more than doubling of resources at 96k). Thanks for your answer. BTW, what dithering do you use? And, what do you use for conversion to 44.1k? (I've heard many suggest BarbaBatch for this).

[Dietz]

Hello David,

as long as you don't plan to do lots of processing like EQs, you won't win much by using 96kHz - in fact it is quite possible that you add unwanted artifacts due to the sampling-rate conversion back to 44.1kHz.

OTOH - it is widely agreed upon sound engineers that digital EQs and filters (and to some extent even digital dynamic processors) provide more pleasing results at higher sampling rates. So - for heavy mixing, it may be a good idea to upsample your project to 96kHz despite the side-effects mentioned above. Be aware that this will put higher strain on your disks and your CPU.

Personally, I tend to stay at the target sampling rate as long as I don't leave the digital domain for a mixdown, but I try to use "upsampling EQs" for critical signals whenever possible (like the Algorithmix' EQs, or the slightly obsolete OctoQ from Steinberg). - For mixing on an analogue board, high sampling rates have their undisputed beauty in the case of delicate acoustic recordings, but they don't make much sense with our Vienna Instruments as they are already delivered in 44.1-format.

... I hope this sheds some light on the issue! [:)]

All the best,

[Dietz]

@Mystr Tiger said:

[...] BTW, what dithering do you use? And, what do you use for conversion to 44.1k? (I've heard many suggest BarbaBatch for this).

Dithering has nothing to do with changing the sample-rate. Dithering is a usual concept for masking rounding-errors when changing the bit-depth of a signal (e.g. going from 24bit to 16 bit for CD-delivery).

[DM33]

If you choose to set your project bit rate to 24bit you will need to dither down to CD for CD burning.

There are multiple dithering algorithms and you should learn about them because it
will be important for you to know which one to use and when.

You can at least learn a little about it here:

http://en.wikipedia.org/wiki/Dither

Each Dithering program will use differing algorithms so you should check documentation.

I use the Waves Masters bundle and Izotope's Ozone. The Waves bundle has the IDR plugin for dithering.

Like Dietz said, if you use the wrong algorithm it will create artifacts.

If you had to choose between an increased sample rate or bit depth I would choose
bit depth.

16 bits will deliver 96db of dynamic range whereas 24 bits will offer 144db. You will get 6db per bit.

The hotter you record the more bits you will use and the more bits you use the more resolution the audio will have.

I think VSL chose to use 24/44.1 for this reason. Of course size was also a consideration but they made a choice to do higher bit depth instead of sample rate.

That says something to me.

Hope that helps.

Thanks,
David Mikautadze

BTW, I keep changing my user name from Dragon33 to my name but it still keeps coming up as Dragon33. It won't save as David Mikautadze.

Anyone have this problem?

[Mystr Tiger]

Much thanks to you both, Dietz and David. Your comments are helpful.

[cm]

david, AFAIK changing the username is not possible with the forum software ... i had to do that directly in the database.
christian

ps: you need to login with your new username now ...

[Nick Batzdorf]

There are four words guaranteed to start a brawl in a room full of mastering engineers. Three of them go together: "stereo miking techniques."

The other one is "dither."

And the differences between different flavors of dither are... well, let's put it this way: I believe in ridiculously expensive cable (within reason); I don't get excited about which kind of dither I'm using.

There, I've said it.

[6]

[DM33]

Ok Nick.

But some dithering algorithms produce audible artifacts when the audio is reprocessed and some other algorithms don't.

So, when I spoke about which dither to use its not because some sound better than others its more about whats going to happen to the audio after its been dithered.

If someone is producing a piece of music and the delivery format is 16/44.1 but is using higher bit depths during the production they should be aware of the fact that after delivering a 16/44.1 final audio file which has been dithered that it could cause artifacts if the mixing engineer decides to run some process on the audio.

Using the right algorithm in the first place could avoid this problem.

If you have some other info you can share to shed some light on the subject I would really love to hear it.

This is the way I have always thought about dither and its use.

I'm not a mastering engineer but I do come in contact with George Massenburg every now and then and will chat with him about it.

I look forward to hearing your comments.

David

[Nick Batzdorf]

Well, I'm the wrong person to ask, because I've never actually listened for the effects of multiple ditherings. I do know that Waves says something to the effect that [I forget which of the types they offer this was - i.e. not Type 1 normal] is designed for multiple ditherings because *in theory* the dithers could "stack up" (not their words) and cause problems...however they've never known that to happen.

In all honesty I can't think of a situation in my work in which this has come up - I've never had any reason to dither twice. I suppose it's possible with the older 16-bit VSL versions, though, since they must have been dithered from the original 24-bit recordings. Knowing VSL, however, I'm sure Dietz et al. thought of that and used the right dither.

In any case, I certainly don't want to be the one to argue that subtleties aren't important! My point is just that there are a few steps before you get to "Which dither do i use." [:)]

And I know George Massenberg will have an informed opinion about dither types, because he's way into this kind of thing.

[Dietz]

@Another User said:

I suppose it's possible with the older 16-bit VSL versions, though, since they must have been dithered from the original 24-bit recordings. Knowing VSL, however, I'm sure Dietz et al. thought of that and used the right dither. [...]

It is no secret that you have to avoid noise-shaped dither like UV22 for signals that will receive further processing down the signal chain. And yes, we took care for that! [:)]

[Nick Batzdorf]

@Another User said:

If you do it by the book, you should always dither as soon you change the volume of a signal.

Well, I should rephrase that and say that I've never had to dither from 24 to 16 bits more than once - i.e. I wasn't talking about processors that dither from higher internal resolution (e.g. 56-bit accumulation) back to 24 bits. Dithering to 24 bits is never a user setting, as far as I know.

[mpower88]

For my 2c thrown in here...

For the fact that 96k is going to more than half your processing power of your machine... I doubt that it is worth it.

Upscaling the VSL samples which are 44.1 to 96 will NOT improve their sound quality. They would only sound better if you got the original 96 *recordings*, which I guess VSL will release when it's technologically appropriate. Mixing 96k samples in 96k makes perfect sense, but upscaling to mix does not except for the point Dietz makes that the EQ's themselves sound better at this higher rate, I haven't tried this but it makes sense to me, anyway he's the expert [;)] the only thing with that is that you need a very powerful machine to do such a heavy mix at 96 so you probably woudn't be able to do a mix of that size at 96k anyway, so it's a bit of a catch 22 unless you run HD3. In that case, just stick to 44.1 - the native rate of the samples.

Miklos.

[Mystr Tiger]

@Dietz said:

If you do it by the book, you should always dither as soon you change the volume of a signal.

Thanks for your comments Dietz. Question: I would think the DAW software would do this dithering internally anytime the gain is changed, true? And, intelligently with respect to how they do the summing (e.g., I think ProTools uses 48 bit summing path, so it seems possible that none of the gain changes require dithering until the final sum is presented back to the 24 bit path). In other words, I would not think that I need to add a dithering plug-in to each of my channels where I have set the gain to something other than unity or zero. Not sure if this is what you meant.

[Nick Batzdorf]

Pro Tools now has a dithered mixer - at least i assume that's the one they're using now. A few years ago there was a public outrcy saying that a dithered mixer was necessary, and Digidesign posted two mixers to see whether people could tell the difference. One dithered back to 24 bits the other didn't.

And I forget whether the results were conclusive, but I believe the one they use now is dithered. (I've been out of TDM world for a while and haven't kept up with this; much as I wanted to justify the upgrade to HD I just couldn't, so I sold my MIX system and for now just have an MBox that serves as a dongle for the PT software when I need it.)