Menu
What's new
By:
Filters

Composing in 96k, Summing at 192k, Bouncing at 48k?

Prockamanisc

Prockamanisc

Senior Member
Does anyone know much about sample rates?

Here's my setup:

Tons of VI's running in Cubase, the project is set at 96k. All tracks get grouped down into pair of stereo outputs, which go out externally into a summing unit. That summing unit goes back into a Burl B2 which is converting at 192k, which then gets recorded back as an audio track in 96k. Then I bounce out what I recorded as the deliverable, and I bounce it out as 48k.

Is this a good way to work? I figure that capturing the analog summing back in at 192k would be beneficial, even though it's being recorded back in at a lower rate. Am I correct in this assumption, or am I missing something?
Click to read more...
 
First time I scored a short film, I sat in the theatre on opening night anxiously awaiting my "moment" in the score - an amazing string crescendo as the camera rose above the trees.
When the moment came, the dubbing guy had buried all my hard work under some tractor engine sound fx.
There's probably a lesson in there somewhere.

You're obviously working at a higher skill level than I am..but seeing as you're asking..it all sounds like a lot of work. Next logical step is to bounce straight to 48 within the DAW, and another version via your mastering chain and see if we can hear the difference.

I smell a 30 page argument incoming..
 
robgb said:
Snarky comment removed...
Just record in 48k. You'll be fine. Most of your targets will likely be listening in lower resolution anyway.
Click to expand...

I agree, to me it seems pointless to work in anything higher...let alone 96k.

@Alex Fraser that is a hilarious story! Almost exactly what happened to me once. My music never seems as "epic" as it does when I'm scoring it, and in the final cut, it's usually at such a low level that it's hardly noticeable in most scenes.
 
Prockamanisc said:
...into a Burl B2 which is converting at 192k, which then gets recorded back as an audio track in 96k
Click to expand...
No, you would either set the Burl ADC sample rate at 192 kHz and record that 192 kHz signal - or set the Burl DC at 96 kHz and record that.

There is no point in using 192 kHz ADC and then resample to 96 kHz using your computer. First, it adds another time consuming step. Second, sample-rate conversions in your DAW may add aliasing and you would end up with worse audio than if your ADC used 96 kHz in the first place.
Also, your 96 kHz score would probably end up in a 48 kHz project anyway so it would be resampled twice.

Finally, which interface is being used by Cubase? Were you thinking of recording the 2 ch mixdown (AES/EBU signal from Burl) into the same interface+computer+Cubase? Or using another interface?
 
Unless you have some original source material recorded at a resolution above 48k, I would think there is no benefit, and certainly no audible benefit, of working at a higher resolution.

I have heard people debate this for a long time. Here's a white paper on it by someone more knowledgeable than most: https://www.lavryengineering.com/pd...the_optimal_sample_rate_for_quality_audio.pdf

It reads, in part:

"...high quality audio converters operating at sample rates no higher than 96 KHz offer results that are very close to the desired theoretical limits. Yet, there are many who subscribe to the false notion that operating above the optimal sample rate can improve the audio. The truth is that there is an optimal sample rate, and that operating above that optimal sample rate compromises the accuracy of audio. To some, this may seem counterintuitive, but is completely proven; whereas most supporters of higher than optimum sample rates offer only subjective results in support." Dan Lavry et al.
 
Prockamanisc said:
Does anyone know much about sample rates?

Here's my setup:

Tons of VI's running in Cubase, the project is set at 96k. All tracks get grouped down into pair of stereo outputs, which go out externally into a summing unit. That summing unit goes back into a Burl B2 which is converting at 192k, which then gets recorded back as an audio track in 96k. Then I bounce out what I recorded as the deliverable, and I bounce it out as 48k.

Is this a good way to work? I figure that capturing the analog summing back in at 192k would be beneficial, even though it's being recorded back in at a lower rate. Am I correct in this assumption, or am I missing something?
Click to expand...

you wont get any benefit. and its overcomplicating things. in audio post production the last thing you want is sample rate issues. keep it at 48k. the only time you might want to have anything more "high fi" would be when you are recording a string session or real instruments and once recorded you converted into 48k 24bit. and THATS even too much.
48k is more than enough.
 
gsilbers said:
you wont get any benefit. and its overcomplicating things. in audio post production the last thing you want is sample rate issues. keep it at 48k. the only time you might want to have anything more "high fi" would be when you are recording a string session or real instruments and once recorded you converted into 48k 24bit. and THATS even too much.
48k is more than enough.
Click to expand...


Not necesarily. Some people set their project rate higher for upsampling plugins (although most processes vulnerable to aliasing should upsample internally anyway). Some also claim that reverbs rendered at higher sample rates sound better. I havent AB that though.
I dont think that high sample rates are necessary. And if you go up dont go down again until the final export.
 
I'd ask why you're using 96k, or 192k? Are you doing heavy sound design work along the way? It might be worth taking a look at why 48k is a standard, even though we can output at 96k very easily and without issue.

48k was landed on (after 44.1 of course) because of the Nyquist frequency (https://en.wikipedia.org/wiki/Nyquist_frequency) - this is generally why so many thing won't need to be output at anything above 48k, since it will cover everything a human can hear and it's corresponding Nyquist frequency. When you're using 96k you're getting Nyquist frequencies but only ones available WAY outside human hearing. It's handy to keep this information before summing if you are planning to do heavy editing to the source audio (ie massive detuning, denoising or whatever for sound effects or sampling, common for those big low "booms"), but again, you only need it on the source audio, the final output doesn't need that information, you'll detune it or do whatever you need to do to it, then sum is down to 48k, still maintaining those Nyquist frequencies now kept within human hearing.

This is just one small look at why 48k is considered "optimal" (and I use that term very loosely as it's always up for debate among audiophiles), but it's a good fundamental look at one of the basic principles of how 48k became a standard.
 
If your VI's are all 48k then why do this? A lot of scores are mixed at 96k (because the studio requires it) but then everything was recorded at 96k and the synth masters are still at 48k.

Why take the huge performance hit? I would just use oversampling on select plugins. Theoretically if you're going out and back in that could help removing any artifacts introduced by the antialiasing filters.

Why not try out the setup at various sample rates and decide for yourself whether its worth doing?
 
Spitfire is recorded at 96k, plugins supposedly work better at 96k, and this video sounds viscerally better at 96k (). Having both played in and conducted real orchestras, it's hard to be completely happy with the sound I'm getting out of my DAW, and this is my latest attempt to close the gap.
 
Prockamanisc said:
Spitfire is recorded at 96k, plugins supposedly work better at 96k, and this video sounds viscerally better at 96k (). Having both played in and conducted real orchestras, it's hard to be completely happy with the sound I'm getting out of my DAW, and this is my latest attempt to close the gap.
Click to expand...

Their output is 48k though. They record at 96k (just like us) but deliver in 48k.
 
Alex Fraser said:
First time I scored a short film, I sat in the theatre on opening night anxiously awaiting my "moment" in the score - an amazing string crescendo as the camera rose above the trees.
When the moment came, the dubbing guy had buried all my hard work under some tractor engine sound fx.
There's probably a lesson in there somewhere.

You're obviously working at a higher skill level than I am..but seeing as you're asking..it all sounds like a lot of work. Next logical step is to bounce straight to 48 within the DAW, and another version via your mastering chain and see if we can hear the difference.

I smell a 30 page argument incoming..
Click to expand...
I had a placement in US tv show, where my music is playing on the radio as an elderly lady and her son are driving in a convertible car with the roof down. They are having a conversation, but after about 10sec the mum says:

"Can we turn the radio off, please"

:crying:
 
There is no scientific reason to work above 44.1. Film is 48 due to convention (easily divisible by 24 frames per second), but going above 48 makes zero sense unless you're going to be pitching it down later, in which case, yes, 96 or whatnot is a good idea.

You're also causing your rig to work *much* harder, so you're losing out on tons of CPU headroom.

Don't work how you are. Work at 48 if you're in film, 44.1 if you're not. End of story.
 
AlexRuger said:
There is no scientific reason to work above 44.1. Film is 48 due to convention (easily divisible by 24 frames per second), but going above 48 makes zero sense unless you're going to be pitching it down later, in which case, yes, 96 or whatnot is a good idea.

You're also causing your rig to work *much* harder, so you're losing out on tons of CPU headroom.

Don't work how you are. Work at 48 if you're in film, 44.1 if you're not. End of story.
Click to expand...

Well if it comes to recording and non linear processing thats not true. The idea behind higher sample rates is to use less steep anti alias filters and thus getting less artifacts out of them.

44.1 Khz is definetly enough to grab the whole audible spectrum (unless you are a super infant and hear even higher than 22k :P ). Thats why it is our common standard. Its just the arbitrary chosen sample rate that covers our whole frequency spectrum. We need more than twice the sample rate to reproduce a given frequency.
 
The only time I ever worked with 96k was on a film where we had sampled rhythmic elements and the composer wanted to radically timestretch & tune them to match different tempos and keys. Even for that, we created a 96k resampling session to bounce out material at the demanded tempo & pitch, and the actual VI sessions were still in 48k.
 
You must log in or register to reply here.
Top Bottom