Originally published 2016
In a recent post I apparently gave out a bad link to this fellow’s video so here’s another. In the video, when you get a chance to watch it, you’ll probably pickup quite a few valuable tidbits of information and get a good view of how an engineer might tend to look at the issue of digital audio. To most engineers it’s pretty cut and dried, as you see, and with respect to the information he’s giving you, who could argue?
I could. Basing a conclusion that high resolution downloads are “silly” on the limited data he gave us is what’s actually silly. This video is a classic version of a smart, well meaning person trying to prove what you know to be true isn’t. To do this, it’s pretty easy. You simply limit the facts you are presenting to a small subset of the whole picture, then spend the entire presentation proving, point by point, the accuracy of your facts. At the end of the presentation, not one fact presented can be argued: ergo the original statement must be true.
Not.
This is the same technique used by a magician trying to trick you. The magician shows you only what he wants you to see and, based on what you see, the impossible happens and you’re baffled. You know that he couldn’t possibly have split a person in half, yet the evidence in front of your eyes is irrefutable. Good stuff. The difference between the video I asked you to watch and a magician is, sadly, the presenter in the video is NOT trying to trick you. He believes everything he is suggesting including his conclusion – which is a shame because we need bright engineering people like Mr. Montgomery helping us, not trying to show us what we know to be true isn’t. That’s the shame of it all. But just for the record, let me suggest that MANY high resolution downloads are no better than the original CD’s they purport to improve upon. That too is a shame, but it doesn’t mean that ALL high resolution downloads are “silly”. Far from it.
Let me suggest an experiment. This is one I perform all the time. Take one of Reference Recording’s fine offerings that are available on both HRX and CD. Play the CD, then play the high resolution HRX disc of the same track. Remember that one is a direct bit for bit copy of the master tape, the other a downsampled version of the same. If you, or Mr. Montgomery, can’t hear the difference then you either need an upgraded system or have your ears examined. The differences are plain, obvious and unmistakable. I am also sure after many conversations with the recordist, Keith Johnson, that the differences are also measurable if you were to simply take the analog outputs of the two identical tracks and view them on a difference scope.
So let’s be clear about this. If you record an original performance at 176kHz/24 bits (as they do at Reference) and then truncate it (downsample) to the 44.1kHz 16 bits of a CD, you will absolutely change much more than just the signal to noise ratio and dynamic information. There’s been many a paper written on this and the evidence to your ears should be overwhelming. Now, if you do the opposite, the opposite doesn’t happen.
If you record and original performance at 44.1kHz/16 bits and then upsample it to 176kHz/24 bits, the differences (if any) will be minute. Further, any differences you might hear probably have more to do with the DAC you’re playing it on than the actual music.
One last point on this video: phase response. Looking at a single 20kHz sine wave and demonstrating it is the same as the input from the generator shows only one aspect of importance to the sound. At a minimum, Mr. Montgomery should show the phase differences between the input sine wave and the output sine wave. It is essential the two be identical and that depends on the A/D converter and the sample rate in use – plus the filter that A/D converter uses to conform to the Nyquist frequency. Can it be the same? Absolutely. But showing us only a small sampling (not to make a pun) of the big picture is really misleading.
I applaud Mark Montgomery for an excellent video. I just wish the title had a different conclusion. Proving to us that what is plainly obvious isn’t true is a good magic trick, but one not appreciated in these quarters.
PAUL McGOWAN / PS AUDIO
What you say about upsampling vs. downsampling and the audibility of differences thereof doesn’t contradict Montgomery’s thesis, viz. audio signals containing inaudibly-high frequency content can nevertheless cause intermodulation distortion products in your amp that contaminate the audible range. Downsampling would erase these products, hence the audible difference, but upsampling would not create them, hence the lack of any audible change in this case.