When Should I Record At 96kHz?

The standard audio CD plays back at a sample rate of 44.1 kHz and a bit depth of 16. As this is the highest resolution the vast majority of listeners will ever experience a recording at, and considering the fact that the upper threshold of human hearing is about 20 kHz, is there any real benefit to running your DAW at 96 kHz? Well, actually, yes there is…

When It Sounds Good Enough

Some people justify the use of 96 kHz recordings on the basis that they just “sound better”. Although we can’t actually “hear” frequencies above 20 kHz, they may still interact with frequencies we can hear, and subtly influence our listening experience.

While this may be true, many producers don’t see it as a compelling argument for recording in a format that takes up more than twice as much hard disk space as the humble 44.1 kHz file – particularly if you’re aiming at the general public rather than the audiophile market.

Relatively few people can actually tell the difference between a 128kbps mp3 and an audio CD under normal listening conditions. Of course, this is not to say that we shouldn’t try to increase the public’s listening abilities; as increased audio quality becomes more prevalent, standards and expectations should accordingly rise.

However, listening tests have been conducted which seem to indicate that even audio professionals cannot reliably tell the difference between 96 kHz recordings and 44.1 kHz recordings… although there are plenty of arguments about the validity of listening tests in the first place.

A Time Stretch Of The Imagination

The merit of 24-bit versus 16-bit recording is technically verifiable; the former has a noise floor of -144 dB, providing a far greater dynamic range than the -96 dB of 16-bit.

I came across a very interesting site recently – The Music of Sound – which provided a clear example of when it is technically advantageous to work with 96 kHz audio (or even higher, for that matter).

To wit, the higher sample rate is of great benefit when timestretching, particularly when slowing down audio. Although many DAWs offer timestretching facilities, a lot of the extreme pitch-shifting relies on interpolation – so if you slow down a piece of audio too much, the program will have to start filling in larger and larger gaps, estimating what should be there.

If you’re working with a higher sample rate, there are more samples per given unit of time – which means that you can slow things down much more before artefacts begin to appear and the audio becomes unusable.