Question about Sample Rate, Processes and Transforms

Hi! I record at 48Khz and I typically set the Spectralayers project sample rate to 48Khz as well. However, It seems that if I set the Project Sample Rate to 192 or 384Khz, processes like EQ Match and Transforms sound much better. Would recording my audio natively at 192Khz provide any additional benefit with respect to these processes? Would I be able to make more extreme edits before a reduction in audio quality is perceived. Thanks in advance! :slight_smile:

What I learned from discussions about sound quality: Something sounds better if you think it does.

Frankly, the subject of optimal sample rates has been beaten to death. You end up with little hard data, and a lot of emotion. A pro engineer I know doesn’t go higher than 24/88.2 for his master recordings, and that’s only when clients demand it. He usually doesn’t go higher than 24/44.1.

1 Like

Thanks for the reply! I know exactly the sorts of discussions you’re talking about. They often seem to involve a lot of black-magic, superstition and $2000 audio cables, lol. My situation is a bit different though and in retrospect I didn’t ask my question very well. I’ve been using Spectralayers to make fairly extreme pitch changes and EQ matches for the purposes of creating interesting new sounds. As I understand it, changing pitch effectively reduces the sample rate, hence my curiosity in working with >48Khz audio. A use-case I’m interested in is taking ultrasonic information and pitching it down into the audible spectrum, which ultimately necessitates higher sample rates for the recording anyway. I just upgraded to a 24/192 interface and have been getting some interesting audio from the >20khz range, which I’m having fun with. Just need a better mic now, lol.

Anyway, I’m answering my own question in real-time through trial and error. Seems like for my purposes, there is benefit to recording at 192 and setting Spectralayers to a high sample rate as well. Thanks again! Cheers! :slight_smile:

1 Like

I’ve heard from folks on both sides when doing restoration and cleaning, that their way is best. At the end of the day it’s what sounds good to you and what your machine can handle. At the end of the day it all gets compressed for streaming,LOL.

1 Like

To add some more input, you also have to be aware that the FFT Size parameter affects how several processes are performed: The Importance of FFT Size

Lowering the FFT Size can have the same effect (in term of processing) as raising the sampling rate. So it’s possible that in some case what you perceive as an increase in quality by increasing the sampling rate can also be achieved by lowering the FFT Size - or increasing it, depending on the context and type of material you’re dealing with.

1 Like

Indeed! I’ve noticed that for EQ match in particular, FFT size is a major variable.

So I ended up ordering a mic that goes well into ultrasonic, so my original question is somewhat moot - as I’ll most definitely be Recording at 192kHz now. I’m very excited about the sound design possibilities! :smiley:

As an aside, does anyone know how well Transform and EQ Match processes benefit from more CPU cores? Or are they mostly single-threaded processes.

They half-benefit from more cores. They are not all fully used, but more than a single core is used.

1 Like

Thank you! Good to know. Processing times are getting tiresome on my ~8 year old i7. I think it’s about time I put the old guy out to pasture. :slight_smile: