“I’ve been considering an upgrade to an audio card that will record at 96 kHz. I’ve heard mixed reports on whether there is any benefit to this. I understand how subtle a difference it might be to have audio recorded about our 20 kHz hearing range. But I’ve also had people tell me that even lower frequency signals can benefit from higher sampling rates. And it does sort of make sense that higher sampling rates would allow finer resolution on all signals, including lower frequency ones. Can you shed any light on this?”This is a bit of a controversial subject. There are many interests trying to push 96k recording forward, but there are also people who claim there is little or no practical difference. For a specific response to your question we polled our resident 96k sales engineer, Nika Aldrich. He says:”There are several reasons that you might want to consider recording, mixing, or effecting your music at 96k. The reason you propose [the sample rate being beneficial to low frequency signals] should not be one – low frequencies aren’t affected by higher sampling rates.””The science behind the A/D and D/A conversion process is actually quite complicated, and filters enter in to the equation in a large way.””In the A/D conversion process, all information above 20 kHz or so is filtered out of the signal. The result is that any high frequency square waves, triangle waves, or other complex waveforms become sinusoidal. Sampling is then done at even increments of time.””If you look at the resultant samples on a computer screen or in some other form, a misrepresentation will be done where, at high frequencies, the samples end up looking like square waves, or other rigid, non-sinusoidal waveforms. The computer software out there does a sort of “connect the dots” in straight lines and gives you the perception that this distorted picture of your original sample is actually what you will end up hearing.””What actually happens is that this information is fed to the D/A converter where filters once again do the process of eliminating all frequencies over 20kHz. The square waves that get fed into the converter get filtered to exactly the same shape that they went in at – at the correct amplitude even. Thus, by the time it leaves the A/D converter it is exactly the same shape and size of waveform that got converted in the beginning.” [Ed- As long as you are dealing with sound captured below the Nyquist frequency it only requires two known points (samples) to be able to properly reconstruct a sine wave.]”For this reason, 44.1k is all that is needed to accurately capture and reproduce all frequencies from 20 Hz to 20 kHz, strictly speaking.”Again, this is not to say that higher sampling rates aren’t valid for recording signals above our known range of hearing. There are a number of studies that indicate we can sometimes perceive differences between material having no ultra high frequency information and material with such information. You might take into consideration that frequencies above 20kHz aren’t so much heard as felt (or perceived, anyway). The range of upper harmonics for a chord played on a piano, for instance, reaches well above 20kHz. The harmonic interplay at these frequencies contributes to the ‘air’ of a recording and has been argued to have psychoacoustic effects as well.And what will be the defining factor in your decision? You will. Listen to what it sounds like to your ears, and make a decision based on what you want for your audio. Do a little research, and decide if it fits into what you want to do.