A premise.
When I was a child, old silent films, which had been shot at usually 16 photograms/sec, were played by 'modern' machines at 24 p/sec. The result is what mostly contributed to our enjoyment and sensation of slapstick comedy's effectiveness: an excessive speed of movement.
Later on, someone 'remastered' some older films interpolating a photogram every two, bringing the total of still images in the film to 24/sec, although one was a simple duplicate of one of the other two. The speed of people's movements became normal, and we lost much of the sense of enjoyment, other than being forced to reset the criteria by which we found old films funny.
Now, let's apply this to sound recording (and let's be patient with my approximated knowledge of technicalities). When we speak of 24/96 (or 24/192, or 32/384 and so on) files, we can presume to be facing one of three options: the original analogue master has been properly remastered and is now available as digital file in various formats/bitrates; the file has been natively recorded at one of those bitrates; it is a simple, oversampled, CD-size 16/44.1.
Oversampling is probably the closest approach of so called HiRes to the cinema example I was doing: a more or less sophisticated (and musically educated!) algorithm interpolates bits where there aren't any, rounding off the scaled, 'stepped' original waveform.
There is an obvious difference between the three procedures, where a native, good digital recording is the best; second, digitalizing analogue tapes. Last, upsampling.
Now, the questions:
– When you buy from download sites music in HiRes, how do you know which of the three options has been used for the piece you've chosen? Do you trust that any would be 'better' than a well recorded and printed 16/44.1?
– Are we sure that we can believe that a site like HDTracks gets original masters? Wouldn't it be safer to suppose they mostly upsample 16/44.1s?
– Do you always get the difference between a RedBook file and a 'HiRes' one? How can you know it's not a mere matter of equalization? Is it always safe and sensible to buy an immaterial file because it bears the tag '24/192'? How much of your enjoyment relies on expectations and trust?
Studies have been done about our ability to tell a sample rate from another, and not infrequently people have not recognized, or preferred, a 24/96 over a 16/44.1. I wonder if buying HiRes files is not often just a way to justify the purchase of a 20,000 streamer.
Just my opinion, thanks for reading.
M.
Posted on: 01 October 2018 by Massimo Bertola
Thanks for the replies. It doesn't seems a minor matter to me, considering the diffusion of so-called HiRes files (I wonder if and how many have, at home, some software able to read a file an tell exactly what it is) and the cost of high-end streamers.
I understand that buying an ND555 can bring to the highest level the experience of even a simple rip of a RedBook CD, but that is more than decently done by an HDX, which has a global quality between CDX2 and CDS3 (not my guess, I have listened and compared a number of times) and can be brought s/h for around 2500/300 (whatever the currency).
I am NOT discussing the choice of basically buying a Squeezebox that, completed with its two power supplies, costs like a good car, but am wondering how buyers of so-called HiRes files are sure of what they are paying for, and most of all of what they're hearing.
Best
M.
Posted on: 01 October 2018 by Mike-B
Hi Max, you ask some interesting questions. I'll skip the underlying premise & go on to what I do & what I have found.
I do systematically check all my HiRes purchases; they are all first downloaded to my laptop & I systematically check with a software program called MusicSCope. Its a quick & easy visual check to be sure it is a true 24-bit & the expected sample rate & that it is not an up-sampled 16/44. Why do I do this ?? - I've been caught out in the past, but I have to say not with the established HiRes & specialist vendors. Because I do this I have now become pretty confident that vendors such as Qobuz, HighResAudio & the very specialised vendors do sell as they advertise.
Judging only by the contained metadata the vendors do seem to get the official record companies HiRes released files - the 'masters' as you call it - How they process those files in house I'm not sure, I expect they do need to convert to the various codec (WAV, FLAC etc)
Is it worth it & how do they sound - the million dollar question. You ask do I always get a difference between RedBook & HiRes ?? Hard to say "always" as in my case an old ripped CD will only get replaced if the album is re-issued with some well regarded remastering. But one recent case I had concerned a London Grammar album "Love is a Beautiful Thing" this is a recent release & I had a 16/44.1 CD rip I had "borrowed" but it had a minor defect & I replaced that with a 24/44.1 download - note the only change was 16-bit to 24-bit - the difference is audible.
Posted on: 01 October 2018 by Innocent Bystander
A fraught subject, especially as Higher resolutions tend to cost more! Some higher res seem to sound better, but they can sound worse, and the absence of definitive information about source is distinctly not consumer friendly.
To put in conrext, I well recall some years ago a friend brought some CDs round and we were having a listening session. We happened to put on one that we had already played, and there was something odd: we both thought it sounded different. Checking more closely we found that between us we had that one CD duplicated - and the two sounded quite different. They were both the same album by the same band with the same release date and will information on the CDs and liner notes identical, apart from one said made in England and the other made in Germany. We could only conclude that there was some unknown difference such as the mastering that made them sound different. (And that put paid forever to any direct comparisons of LP versus CD!). Whether one or other spunded better wasn’t something we considerered - but a matter of preference as to which version one preferred, und unsurprisingly perhaps in that session we each preferred our own, being the version with which we were familiar.
There is one website I know of where you can compare different resolutions: 2L records (2L.no).They have an area on their website called test bench, where they have a number of music samples from the same source but in different resolutions that can be downloaded to compare. (Something I keep meaning to do that haven’t got round to...)
Posted on: 01 October 2018 by Huge
I check HiRes files using Audacity rather than MusicScope (I'm a cheapskate!), and I have found the same as Mike-B.
I can also add Linn to the list of trustable sources, and they do some exceptionally fine recordings particularly with the SCO, such as the Weber Wind Concerti and Berioz' Symphonie Fantastique (SCO, Ticciati) for these and some others the benefit of 24/192 is very clear to me.
I've experimented doing double blind tests with some friends and with these two pieces they preferred the 24/192 encoding and easily distinguished it (100%) from 16/44.1 or 16/48 recodings of the same files. 24/48 was distinguished most of the time, but not always quite such a clear result.
