Everything you always wanted to know about sample rates and cab simulation Impulse Responses

What sample rate should you choose for your IR?

Impulse Responses for cabinet simulation are often offered in sample rate flavours like 44.1, 48, 88.2 and 96 kHz. What sample rate should you choose?

Well…

Let’s test it!

I created a 22 kHz sine wave in Audacity and exported it as a 24 bit / 96 kHz file. I then imported the file into Ableton Live and used the Ignite Amps NadIR as impulse loader. The Ableton Spectrum device was used for analysing the frequency spectrum.

What happens when we are working in 44.1 and we use an 96 kHz IR?

In Ableton Live under Preference I set the Sample Rate to 44100, this means my soundcard will be using this setting. On Spectrum however the 22 kHz frequency still shows up. This is a bit confusing because what we see we cannot hear. But since Ableton is set to 44100 my soundcard will output none of the frequencies above 20 kHz.

To make sure what’s going on, I make an export at 24 bit 44.1 kHz. When I open this file it contains silence. Which makes perfect sense, since the 22 kHz frequency shouldn’t be there.

Conclusion 1: using a 96 kHz IR on a project using a sample frequency which is lower makes no sense. Any of the frequencies above 20 kHz the 96 kHz IR might contain will be ignored. So the 96 kHz IR file is working as if it is a 44.1 file. Note: your IR software or hardware device will automatically resample. Using a 96 kHz feels like overkill here.

Let’s reverse this, let’s set Ableton to 96 kHz and use a 44.1 kHz IR file.

In Ableton Live under Preference I set the Sample Rate to 96000. On the Spectrum again the 22 kHz frequency lights up. How is the 44.1 kHz IR file able to reproduce this? Well it isn’t. What is happening here is that the 22 kHz frequency is not being affected by the IR, it’s send straight through as if the IR-effect was bypassed for those extreme high frequencies.

Conclusion 2: using an IR with a higher sampling frequency than the project is set to allows frequencies above 20 kHz to be unaffected by the IR-loader. This might result in a signal containing the frequencies convoluted by the IR as well as the unaffected frequencies above 20 kHz (the amp signal before IR-processing). This might result in a brighter signal compared to its 44.1 alternative. But is this a problem? We cannot hear anything above 20 kHz. So no one in the universe should be able to hear any differences between a 44.1 and 96 kHz IR. In theory it might be brighter but we cannot find out by using our own ears.

So 44.1 kHz is perfectly fine as an IR file format for cabinet simulation.

What about annoying sample rate aliasing issues?

A project at 44.1 kHz can’t process frequencies above 20 kHz. But when saturating a signal the harmonics this generates can often be pushed above this 20 kHz limit resulting in aliasing. Most plugins are using oversampling to protect this from happening.

But what about IR’s? IR-loaders will never cause aliasing because they are 100% lineair, they are not able to saturate or compress the signal. An IR only changes the loudness and the ringing/decay of the frequencies.

My tests also proves this. Running 22 kHz through a 44.1 kHz IR file doesn’t cause aliasing frequencies. When running pure since waves through an IR-loader this will never result in harmonics or aliasing frequencies. Ain’t gonna happen, ever.

Conclusion: Low pass is your friend

The fantastic IR-loader NadIR opens by default with the Low Pass set at 6 kHz. Which makes perfect sense because it filters out the higher frequencies which can make an electric guitar sound overly bright. You can set it a little higher when the IR contains a very bright sound, but I wouldn’t recommend you to set it above 10 kHz.

Final conclusion: in my opinion using 44.1 kHz IR files makes most sense. They are small and perfectly fine to create superb sounding cabinet tones no matter in which format your project is set.

In other words: keep it simple. 44.1 is king.

3 Comments

Seems your rule of thumb regarding sample rates may not generalize to the case of a guitar body IR, due to the higher frequency content? (as opposed to a cabinet IR)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.