High Sampling Rates – Is there a Sonic Benefit?

Date:2020/3/3 11:28:10 Hits:

There’s much debate whether sample rates higher than 44.1kHz, like 96kHz or even 192kHz, yield a significant sonic improvement. According to many people, higher sample rates aren’t necessary. Yet others insist higher sample rates are audibly better. So who’s right?

Spoiler alert: Both are — and if you don’t think you need to read further, listen to the audio example.

I’ve recorded quite a few classical projects at 96kHz. In my experience, with audio recorded at 96kHz, no one can reliably tell the difference between that audio being played back at 96kHz or sample-rate-converted down to 44.1kHz and played back. However, I’ve also found that there can be a significant difference between audio recorded at 44.1kHz or 96kHz, and even at 96kHz or 192kHz.

The audio example plays two measures of a synth clav (pulse wave-based) preset rendered at 44.1kHz (rendering is the process of generating a final digital output from a file and is performed in DAWs by exporting or bouncing the file to create an audio file), two measures rendered at 96kHz, and two measures rendered at 192kHz; all of them are playing back in a 44.1kHz project. After listening, I think you’ll agree that the difference is not subtle at all.

As to why the difference is so dramatic, if a signal source like a virtual instrument or amp sim generates harmonics that interfere with the project’s sampling frequency, a certain type of distortion, called foldover distortion or aliasing, can occur. This doesn’t happen when recording audio sources that feed into an audio interface, because the interface has filters to keep these harmonics from entering the computer. But sound generated inside the computer is a different story.

It’s theory time: A digital system can represent audio accurately at frequencies lower than half the sampling rate (called the Nyquist limit — e.g., 22.05kHz in a 44.1kHz project). If a virtual instrument plug-in generates harmonic content above this limit — for example, at 40kHz — then you won’t hear a 40kHz tone, but you’ll hear the aliasing created when this tone “folds down” below the clock frequency (to 4.1kHz, in this case). Aliasing thus appears within the audible range but is harmonically unrelated to the original signal and generally sounds pretty ugly.

How Bad Is It?
Fortunately, this isn’t a common problem; the issue crops up mostly with older virtual instruments and amp sims. Not all plug-ins exhibit foldover distortion, for one of four reasons:

The audio isn’t rich enough in harmonics (or distorted enough) to generate frequencies high enough to interfere with the project’s clock signal.
The plug-in or the host itself can oversample internally, which means that as far as the plug-in is concerned, the sample rate is higher than that of the project. So any foldover distortion occurs outside the audio range.
The project sample rate is high enough to provide the same kind of environment as oversampling.
The plug-in designers have included appropriate anti-alias filtering.
Many modern virtual instruments and amp sims oversample, or at least have an option to do so, so you’d think that might be the end of it — and most of the time, it is. However, there can be limitations with oversampling, especially for virtual instruments.

Although some instruments may perform 2x oversampling, that still might not be sufficient to eliminate aliasing on harmonically rich sources like pulse waves — so oversampling an oversampled instrument can still improve the sound. Furthermore, the sound quality of plug-ins that oversample will be influenced by the quality of the sample-rate conversion algorithms. They have to do their work in real time; whereas, offline solutions aren’t subject to that constraint.

Oversampling also draws more CPU power. Fortunately, with software that lets you choose whether the plug-in oversamples or not, if your CPU is struggling, you can turn off oversampling while recording for minimum latency, and then turn it on during mixing when latency isn’t as much of an issue.

Darn. So Now I Need to Record at High Sample Rates?
Not necessarily, because we’ll describe how to obtain the benefits of high-sample-rate recording at lower sample rates. And that’s good, because recording at higher sample rates has some disadvantages:

96kHz audio takes up over twice as much memory as 44.1kHz audio.
Running at 96kHz stresses out the computer more and reduces the potential track count.
It may not make any sonic difference anyway.
I usually record at 44.1kHz with 24-bit resolution, so the computer runs smoothly and projects don’t need as much memory for storage (if clients want to work at other sample rates, of course I accommodate them). However, with instruments and processors that benefit from being recorded at 96kHz, you can use the following workarounds to derive the benefits of higher sample rates in lower sample rate projects by doing upsampling.

Note that upsampling can’t fix audio that already has aliasing distortion. Upsampling audio to 96kHz that already contains foldover distortion will simply reproduce the existing distortion. This technique applies only to audio created in the computer. Similarly, it’s unlikely that upsampling sounds recorded via a computer’s audio interface will yield any benefits, because the audio interface itself will have already band-limited the signal’s frequency range, so there will be no harmonics to interfere with the clock frequency. However, if someone thinks recording a virtual instrument or some signal processors at 96kHz sounds better than recording at 44.1kHz, that may indeed be true — and as you’ve heard, the difference can be obvious.

Solution 1: Change the Sample Rate Temporarily
If a project contains only virtual instruments and no audio, change the project sample rate temporarily to a higher rate, like 96kHz or higher. If you can adjust your audio interface to the new sample rate, then simply render those instrument tracks that benefit from a higher sample rate. This converts them into audio at the higher sample rate, so there won’t be any foldover distortion. After rendering these sounds, you can then change the project sample rate back down to your original project rate of 44.1kHz or 48kHz. Yes, it’s that simple.

You don’t lose the benefits of upsampling when you later convert the sample rate back down to 44.1kHz, because rendering at the higher sample rate eliminates any foldover distortion in the audio range — and 44.1kHz has no problem playing back sounds in the audio range.

Solution 2: Create a Temporary Project
If audio has already been recorded, you won’t be able to change the song’s sample rate easily, if at all. So here’s a foolproof method if Solution 1 isn’t possible. We’ll assume a 44.1kHz project sampling rate, and that the virtual instrument’s MIDI track has been finalized, but you haven’t rendered it to audio yet. Here’s how to “upsample” virtual instruments.

Save the virtual instrument preset so you can call it up in step 6.
Make sure the MIDI clip that drives the instrument starts at the project beginning. If needed, add space or bogus data to extend the clip’s beginning.
Export the MIDI clip driving the instrument.
Close the existing project.
Create a new project at a higher sample rate, like 96kHz or higher.
Insert the virtual instrument you used previously, and then load its preset.
Import the MIDI clip, start it at the beginning, and assign it to the instrument track.
Render the instrument track starting at the beginning of the song.
Export the rendered audio, then close the project.
Open the original project with the 44.1kHz sample rate.
Import the rendered audio; your recording software should convert the sample rate back down to 44.1kHz when you import it. Use this as the instrument audio instead of what would have resulted from rendering the instrument in the 44.1kHz project.
With amp sims, the procedure is similar except that you save the amp sim preset, extend the audio file driving the sim to the project beginning, and then render in a higher-sample-rate project.

Final Thoughts
Note that you may not want the difference caused by recording an instrument or amp sim at a higher sample rate, or upsampling it. When I did an upsampling demo at a seminar with a particular synthesizer, most people preferred the sound with the aliasing because the upsampled sound was brighter than what they expected. However, when I did upsampling with an amp sim, and also with a different synth, the consensus was that the upsampled versions sounded much better.

Although upsampling isn’t a panacea, dismiss it at your own peril. With many synths, upsampling will make no audible difference — but with others, you’ll hear more clarity and sometimes, a very different sound. As always, use your ears to decide what sounds best with your music.

The point is, now you have a choice to hear the instrument the way it’s supposed to be heard to decide if you like that better, or to leave it as is. After all, distortion isn’t necessarily that horrible — think of how many guitar players wouldn’t have a career without it!

Prev:Stage Lighting 101, Part 1: Understanding the Basics

Next:High Sampling Rates – Is there a Sonic Benefit?

Leave a message

Message List

Comments Loading...

Products Category

Products Tags

Fmuser Sites

High Sampling Rates – Is there a Sonic Benefit?

Leave a message

Message List