Sometimes I get in mixes which have been printed/exported at 88.2KHz. It seems some people believe that working at 88.2KHz somehow will translate better once converted down to 44.1KHz. The reason behind this is because 88.2K it’s a multiple integer number (x2), and therefore, once down-sampled, no approximation will be needed.
This is rooted in the early days of digital audio, where it was common practice to pass a final master through high-end A/D (such as Pacific Microsonics) and convert it down to 44.1KHz in real-time.
Today, with powerful computers and a plethora of high-quality software SRC, the truth, unfortunately, is different and the mathematics involved in both conversions from either 88.2KHz and 96KHz is exactly the same.
But this is nothing new.
Already towards the end of the Millenium, JA Moorer (a digital audio expert who collaborated often for Sonic Solutions) wrote this important research for the AES (Audio Engineering Society) where he analised this very problematic in big depths.
His conclusions were that, subject to the quality of the algorithm of the SRC, there was absolutely no reason to expect an 88.2k->44.k conversion to sound any better than the equivalent from 96k.
Fact is, today’s SRC are all working following the same principle. The original audio is normally oversampled at much higher frequencies shifting the anti-alias filter in the ultrasonic range before it gets downsampled to 44.1KHz.
I have specifically mentioned the quality of the SRC as a big contributing factor to the final sound quality because, as you can see from this chart, SRC DO NOT sound the same and some are much better than others.
So, what is wrong with working at 88.2KHz?
To summarise, there is nothing wrong to work at higher SR than the CD format. But it is deceiving choosing to work at 88.2KHz because of problematics that no longer have any impact on today’s workflow.
MOREOVER… it could be completely redundant if the final delivery has to be for Apple MFiT (which requires a 2496 delivery).
As always, only the final result counts. So, if you like working at 88.2KHz and your mixes sound awesome, no need to change your workflow because of this mathematical conundrum.