44.1 kHz or 48 kHz?

[jamcat]

Most sample packs still in 44.1 kHz though, which is a pity.

This is one of a few reasons why I try to avoid samples, and use physical modeling instead whenever I can.

[billcarroll]

Most sample packs still in 44.1 kHz though, which is a pity.

This is one of a few reasons why I try to avoid samples, and use physical modeling instead whenever I can.

I convert all my samples to from 44.1 to 48 kHz without any issue. I use iZotope RX Resample, but you can also use the free Voxengo r8brain which is https://www.voxengo.com/product/r8brain/

[jamcat]

Most sample packs still in 44.1 kHz though, which is a pity.

This is one of a few reasons why I try to avoid samples, and use physical modeling instead whenever I can.

I convert all my samples to from 44.1 to 48 kHz without any issue. I use iZotope RX Resample, but you can also use the free Voxengo r8brain which is https://www.voxengo.com/product/r8brain/

When I use samples, they're in a VSTi: MODO DRUM cymbals, EW Choirs, the occasional Kontakt instrument.

If I was using loose recordings/loops made by someone else, I would upsample to 48kHz in Acon Digital Acoustica. (It has a very good SRC, up there with the best.)

I always record at 96kHz, though, to bypass the audio interface's Nyquist filter. All Nyquist filters distort the phase all the way down to about 10kHz, no matter how good they are. That's just the nature of steep analog filters. But the filter is only used for 44.1kHz and 48kHz, and is not needed or engaged for 96kHz.

In Studio One, it doesn't really matter what the source sample rate is. It resamples on the fly to whatever your song is set to, and saves CPU during realtime mixing. Studio One's SRC is average, so I wouldn't use it for a destructive sample rate conversion, or mixdown at anything other than 96kHz. Then I'll use Acon to convert the master to 48kHz or 44.1kHz, whichever is needed.

[El°HYM]

Most sample packs still in 44.1 kHz though, which is a pity.

This is one of a few reasons why I try to avoid samples, and use physical modeling instead whenever I can.

There are a few in 96 kHz but also rare and mostly then in the field recording department. It would make life a little easier if the 'new' standard would be 48 kHz instead of the good old CD 44.1 kHz for common sample libraries.

[mixyguy2]

I tried 44, 48, 96 and couldn't hear the difference.

Because there isn't one, other than a tiny latency gain at 96 which also won't be noticeable about oh 99% of the time at least. But good luck convincing the "more is better" groupies.

Cue the Spinal Tap "11" scene lol

[mixyguy2]

I always record at 96kHz, though, to bypass the audio interface's Nyquist filter.

:facepalm:

[jamcat]

I always record at 96kHz, though, to bypass the audio interface's Nyquist filter.

:facepalm:

Do you have some thoughts to share?

[osiris]

I did it. I switched the rate up to 4800 in the Audio settings and then bounced the track at 24 bit. What I found out is: it didn't mess with timings. I can't understand why this happens when I try to upconvert a simple loop in an wave editor. The audio was much cleaner, so if there's any mistakes in edits they will show up more audible. Anyway, hooray.

[ampetrosillo]

I always record at 96kHz, though, to bypass the audio interface's Nyquist filter.

:facepalm:

Do you have some thoughts to share?

Apart from the fact that you don't really hear phase, so it's inconsequential, but unless you have definite information on the matter, it is likely that your converters are actually sampling at something like 1-2Mhz and then downsampled digitally. It is true that antialias filters either distort phase (IIR) or time (o pre-ringing, linear phase FIR), but that's inevitable.

[jamcat]

Phase distortion is not inevitable if you record at 96kHz so the Nyquist filter isn’t active, then resample in software with a linear phase filter.

You’re talking about oversampling, but that doesn’t eliminate the need for a filter, it just kicks it down the road a little bit, from in front of the DAC to behind it. There still needs to be a Nyquist filter (or “decimation” filter) before it becomes 44.1/48 kHz, whether it’s analog or digital. It’s still going to be a high order filter with a ripple that affects about an octave of the audible range below it. Even though it’s digital, it’s still not linear phase, because that would introduce too much latency for real-time recording.

The resampling filter you get in good audio software (iZotope, Acon, Wavelab, r8brain PRO) is going to be much better than the one employed by your ADC.

You may “not really hear phase” but you hear the cumulative effects of it as brittleness in high frequencies. If it can be avoided (which it can) then it should be, which is what recording at 96kHz then converting with a high quality SRC at the very end of the process does.

[Ploki]

Nyquist filter is always present, it’s just set higher.
filtering
Filtering with linearphase comes with its own set of problems.
I have all my voxengo plugins set to mininal phase oversampling. The sound of linearphase irks me

[MidnightRunner]

Most sample packs still in 44.1 kHz though, which is a pity.

i use only samples. this is why i just use 44.1 settings. i tested and i cant see a audible benefit to using anything higher.

[jamcat]

Nyquist filter is always present, it’s just set higher.

I had thought so, too, but from what I’ve read, there is no filter at all with 96kHz, because no audio equipment is passing audio above 48kHz.

But even if it is merely higher, at ~48kHz, the ripple distortion would not reach down to the audible range. So either way, your high frequencies are being affected, which means no brittleness.

Filtering with linearphase comes with its own set of problems.
I have all my voxengo plugins set to mininal phase oversampling.

Yeah, minphase is an option if you’re using r8brain PRO as your sample rate converter. It’s pretty expensive though, but the conversion is supposed to be like an ideal ADC, which will definitely be better than what’s in your prosumer audio device.

It’s also finally available for Mac and is Apple Silicon native.

[El°HYM]

I had thought so, too, but from what I’ve read, there is no filter at all with 96kHz, because no audio equipment is passing audio above 48kHz.

In the end of the day there is no perfect solution with all of this, just workarounds and choosing the most non - destructive way.


https://vladgsound.wordpress.com/2014/1 ... ha-version

[BertKoor]

> no audio equipment is passing audio above 48kHz.

1. My first Sony hifi amp ca 1987 went up to 120kHz
2. Aliasing can get wrapped down back into audible range