The Upsampling Your Mix Thread

[Sascha Franck]

Yes that makes sense, that part I understand, it seems logical and obvious that applying FX on an 'upsampled' version would yield better results than a non upsampled one. BETTER STILL, I would have thought, would be applying FX on audio that was RENDERED at the higher samplerate, AND applied with the higher samplerate?

Of course. But as our current branch of machines isn't capable of dealing with 96kHz mixes, one needs to make some compromises (I don't, I just work at 44.1 all throughout, but I'm not delivering final results to anywhom either), and there's quite some options to do so, such as working in 44.1 and then rendering out some submixes, reimporting them into a 96k project and applying the final FX there.
Kingston already gave a nixe example: Imagine having some drums that are all mixed separately but then fed into a drumbus with some additional FX applied straight on that bus (such as EQs and compressors). You could know just render out this (stereo-) bus without the bus-FX and apply them on the 96k file.

Whether an how much any such things improve your mixes is depending on the used FX. And of course, a shitty mix won't benefit much. The latter being the main reason why I don't bother with it

[Dayl]

Cool, well I am about to get a bunch of raw files rendered at 16/44.1 should I look into upsampling to 96khz before importing into a mix? and then add my eqs etc and mix down?

or should I mix down then export or get the whole track rendered to 96khz? then import the track for some mild tweakage?

What do you guys think? This is the first time I will be mixing a band so I need to impress some people.

[Sascha Franck]

Cool, well I am about to get a bunch of raw files rendered at 16/44.1 should I look into upsampling to 96khz before importing into a mix? and then add my eqs etc and mix down?

or should I mix down then export or get the whole track rendered to 96khz? then import the track for some mild tweakage?

I'd say it just depends.
Before I would start with any such things, I would fool around with a single track (so that your computer can handle it) at 96kHz and try out the various FX you are using.
In case there's clearly noticeable differences with FX that you'd usually apply to individual tracks, the first method would be the way to go. In case it only applies to your mastering FX, the second method would be it.

And of course there's all kinds of "inbetween" scenarios, such as rendering out, say, a choire bus, a drum bus, a guitar bus, etc.

[Sam@Megablastic]

A very good and simple way to demonstrate the 44/96khz thing is to either:

1. Get FM7 synth and play a really high note at 44 then at 96 and just hear the difference

2. Get a full track and boost 12khz freq by 5db at 44 then at 96khz and hear the difference.

Now this is for just one vsti/track, imagine all those errors and harshness that 44khz adds multiplied by how many tracks/vsti's you are using in a mix.

Great topic guys, glad the trolls left.

[Dayl]

Nice one guys, as ever..more food for thought.

[Lagrange]

Cool, well I am about to get a bunch of raw files rendered at 16/44.1 should I look into upsampling to 96khz before importing into a mix? and then add my eqs etc and mix down?

or should I mix down then export or get the whole track rendered to 96khz? then import the track for some mild tweakage?

What do you guys think? This is the first time I will be mixing a band so I need to impress some people.

The cool thing about doing this is you can use something like R8brain or AudioMove to convert pretty much anywhere in your process (ofcourse before you apply hi end effects). I did not catch onto what Kingston and Bduffy were saying until I tried it myself and wow there was a clear difference when using certain reverbs, parametrics, flange/chorus, overdrive effects and synths (those are all I've tryed so far). It just take a bit of pre planning as to where and how to use this technique in the mix I guess..

Still waiting for some kind of feedback on upsampling FFT conversions (for timestreching). God knows I'd love to be able to get it to sound better ..

L

[bool]

Rendering vsti tracks at higher samplerate and mixing rendered stems is a very old, ubiquitous trick. I wonder how can somebody not know it or fight the simple logic behind it.

[jackson]

So I went home last night to test a few of the pointers on here. First thing I noticed when I simply changed the samplerate of a finished track was that some of the VST effects' sounds changed drastically. The most obvious with Prosoniq's Orange Vocoder which, since it's so old, I'm hardly surprised by. I would imagine what I'll do is render that track at 44k and reimport it into the new high samplerate project.

Second thing I noticed was a huge difference in CPU consumption between 88k and 96k. I'm running Live 6 on a brand new Core2Duo E6600 with 2 gigs of RAM, and the project I tested above ran at about 28% at 44k, ~55% @ 88k but slamming 100% at 96k. I wonder if this is again due to an incompatible plugin? Didn't have time to test.

LaGrange was asking about running a vocoder at 96Khz; I tested a couple last night, and the sound was drastically altered by the rate change. Much less mud, way more high end clarity Very different sound though, so you can't expect to just render the track at a higher rate for a better quality sound.

[Kingston]

First thing I noticed when I simply changed the samplerate of a finished track was that some of the VST effects' sounds changed drastically.

I've been doing this long enough that I've pretty much managed to weed out all effects that can't handle variable sample rates with class.

I still find it a bit odd how much trouble sample rate scaling seems to be to awfully many developers.

[kylen]

...I wonder how can somebody not know it or fight the simple logic behind it.

[jackson]

Of course there's another non-audio benefit to running at a higher samplerate.

Latency/2

[Lagrange]

LaGrange was asking about running a vocoder at 96Khz; I tested a couple last night, and the sound was drastically altered by the rate change. Much less mud, way more high end clarity Very different sound though, so you can't expect to just render the track at a higher rate for a better quality sound.

Care to share some names of best/worse vocoders to use @ higher khz? So your saying render @ higher rate and then use vocoder in 96khz project? Or does it matter what is being fed into the vocoder (speed wise)?

L

[jackson]

LaGrange was asking about running a vocoder at 96Khz; I tested a couple last night, and the sound was drastically altered by the rate change. Much less mud, way more high end clarity Very different sound though, so you can't expect to just render the track at a higher rate for a better quality sound.

Care to share some names of best/worse vocoders to use @ higher khz? So your saying render @ higher rate and then use vocoder in 96khz project? Or does it matter what is being fed into the vocoder (speed wise)?

L

I could put together some samples tonight. I stopped testing last night when my mouse decided to kick the bucket

Do you have any free or non-crippled demo vocoders in mind you'd like to see tested?

[Lagrange]

jackson I forgot to mention that the strain upsampling will put on your system WILL attack mouse drivers and circetry..

L

[jackson]

jackson I forgot to mention that the strain upsampling will put on your system WILL attack mouse drivers and circetry..

Maybe it was the extra high frequency content present by running at 96Khz interfering with the mouse's wireless connection