Losing things

It's not always bad to lose things. Take weight for example. Most of us could stand to lose a few pounds, certainly I could. But when is it ok to lose musical data? Is it ever ok? In yesterday's post I mentioned that lossy compression schemes in digital audio are used to make file sizes smaller. But that size reduction often comes at a price, lost data. And once the data is lost it can never be recovered. Take MQA for example. This often talked about format, currently the darling of some in the Audiophile press, is a lossy format. It is reported that what MQA gives up isn't needed anyway, and there's good evidence they may be right, at least intellectually speaking. From what I've read in the magazines the data the MQA process loses is high frequency information outside our abilities to hear. For example, classic Red Book CD covers the range of accepted human hearing: 20Hz to 20kHz, and nothing more. Higher sample rate files, such a 96kHz, 176kHz, and even 192kHz are capable of reproducing frequencies that exceed accepted human hearing limitations. And no one will argue that we cannot hear 50kHz. We simply can't. So, when the MQA encoder throws out the high frequency data we cannot hear, and uses that same space to pack in data we can hear, life should be good. Right? Well, that's the question and one I am not technically able to answer. But here's what I can tell you. An analog preamp or power amp, for example, nearly always sounds better when its bandwidth exceeds 50kHz. I can clearly hear the difference on the BHK preamp between bandwidth limiting at 50kHz and its actual bandwidth of 100kHz. Yet, I cannot hear those frequencies. Reasonable explanations for this mystery include phase shift caused by high frequency filtering, phase shifts in the frequencies we CAN hear, as well as ultrasonics we cannot hear–yet seem to be able to perceive somehow. There's been a number of studies of musical instruments that output harmonics well above our hearing, the most famous of which is by James Boyk of Caltech. Without question when lossy files give up musical data in the audible regions we hear that loss. The more interesting question is what happens when we lose data we don't officially hear?