Discussion in 'Music Corner' started by Steve Hoffman, Nov 30, 2007.
Correct and yes.
We did not master them separately. We used a split feed.
Right, I meant to write that but forgot the term.
Saved another pass on the old tape this way, among other reasons..
@Steve Hoffman , I hope this isn't considered off topic, but what do you think about the recent Mobile Fidelity news?
I read the thread today, oh boy
About a lucky man who made the video
And though the news was rather sad,
I'm now not spending bank.
Now I know what steps are used to master in DSD444444444444444...
I mostly listen to Rock CDs such as Robin Trower, Led Zeppelin and Pink Floyd. I would like Steve to do a comparison on such a music style rather than a 60s jazz recording and give us his opinion on this.
Actually we now know that frequencies above the threshold do affect human perception of music. It was demonstrated using a CT scanner to show that more areas of the brain "light up" when audio playback includes those 22k+ freq.s . So, the Science says, if we sample at 96k we get a more pleasing sounding recording
I was always curious about that possibility. Can you please provide a source for the research on this matter?
first a paper that discusses ultrasonics https://www.cco.caltech.edu/~boyk/spectra/spectra.htm
Oohashi et al. claims that reproduced sound above 26 kHz "induces activation of alpha-EEG (electroencephalogram) rhythms that persist in the absence of high frequency stimulation, and can affect perception of sound quality
This is a short and interesting read (the second one, that is, the first one won't open in my browser due to a security warning), thanks for the link.
But you will have to ask yourself if this bears any significance for our home listening.
1. Do the recordings we listen to contain frequencies below 22 Hz and above 22 kHz in a significant amount to set off these things in our brains? If you listen to a lot of music that has been originally recorded either to magnetic tape or let's say between the 50s and 70s, you have your answer already.
Edit: I misread the article in one point, so the "ultra low frequency argument is void". Sorry. The part about ultra high frequencies stands as it is.
2. Do your speakers reproduce both these frequency extremes? As a reminder, the article states (highlighting by me): "Nevertheless, the power spectra of the alpha frequency range of the spontaneous electroencephalogram (alpha-EEG) recorded from the occipital region increased with statistical significance when the subjects were exposed to sound containing both an HFC and an LFC, compared with an otherwise identical sound from which the HFC was removed (i.e., LFC alone). In contrast, compared with the baseline, no enhancement of alpha-EEG was evident when either an HFC or an LFC was presented separately."
So you need both, really low bass down below 22 Hz, which most speakers/listening rooms simply cannot reproduce, and ultrasonics, which most DACs will filter out anyway, no matter if it is a 192 kHz file you play through it. Low pass filters are mandatory in any DAC, especially when it comes to DSD, because of the so-called "noise shaping" that a DSD converter does to get rid of the noise around 30 kHz caused by the 1-bit-technology of Direct Stream Digital (very abbreviated explanation, sorry). That leads to the third question:
3. Does your Hi-Res file contain unwanted frequencies above the audible range, like noise? This seems to happen quite regularly in consumer audio, or at least it's not uncommon, and if that is the case, do you want your DAC to filter out these frequencies rather than having your speakers struggling with reproducing what is mainly non-musical digital noise?
And, like I said above, almost any DAC does high frequency filtering, either via a brickwall filter (as it is common in 16 bit / 44.1 kHz PCM encoding) or more gentle. But something like 30 kHz almost never reaches your amplifier or your speakers, I would argue.
These are just a couple of thoughts after one read of this very short article. I can't (and certainly won't) discuss the scientifical approach or the quality of the study in any way, partly because there is no information about this in this short summary (at least not that I could find).
But like with a lot of these arguments about the benefits/non-benefits of hi-resolution audio, I find myself thinking about it for a while and almost always coming to the conclusion that there are so many factors involved in my listening environment, the music I listen to and its sources (!), my everyday enjoyment of reproduced music (or lack of ) that simply limit the possibility of any beneficial components of hi-res compared to either CD or LP. As much as I love both formats, both CD and LP have their limitations (in the case of the latter it is severe 2nd and 3rd order harmonic distortion, which apparently a lot of people dig - I guess me included sometimes). But these limitations don't bother me really, or I should say, I don't think they limit my listening experience too bad.
Don't get me wrong, I really like the scientifical research and discussion about this subject, but I am yet to find any real relevance of it for home audio.
Here's another article (more of a summary of existing studies and their value) about the topic. I found it very interesting and educated, but also struggled to find significance for me personally in there.
High Resolution Audio: Does it matter?
I have a copy of the paper in full somewhere. What you are reading is merely the abstract. You can easily download the full paper if you join researchgate
please note, it is not referring to freq.s below 22Hz so absolutely nothing to do with ultra-low bass
The two components are: an audible low-frequency component (LFC) below 22 kHz ( that's the normal human frequency response)and an HFC above 22 kHz
This is significant, and I completely agree. I think it's funny that while it's mostly the audiophile all-analog crowd that claims ultrasonics are important, they are most likely only present on high-res and high sample rate digital recordings.
Theoretically, it might be possible to have 'better' copies of recordings, even from deep history... At least, something to look forward to, hopefully before the end of my days...
Much of the time, we are still apparently not getting the 'full technical quality' of the mix.
I know that the information below can be disagreed with, create discomfort, or be information for an open mind, but there is existence proof of either an ingenius expander scheme (I am not that bright) or an effective decoding of s beautifully stealth 'processing complex' design as done by a genius (there are the characteristics of an R Dolby design in the processing.) There are too many incredibly brilliant ideas in the encoding process that I am definitely not 'brilliant' enough to have invented. After working on the project for many years, the results from the 'existence prover' have started becoming 'artifact free'.
Our personal colllection of recordings, when compared with what was likely 'mixed', there is still usually more hiss, more modulation fog, and other impairments in many normally distributed recordings. Consumer recordings often produce results that are tolerable even to a picky audiophile, but often still not the ultimate quality.
For fun, just directly mic a subject, even record it digitally, don't even worry about the strong dynamics -- listen how clean, pure the locally produced signal is. Then, listen for things like subtle undulations in a vocal, compare with typical consumer recordings. Much of the time, there is a noticeable, if not profound loss of subtle detail. With pure digital transfers, there is no excuse for mistaken subtle distortions added to the recordings, because even we (the consumer) can do transfers practically-perfectly yourself. There is no reason why 1M$ studios and distributors cannot do it perfectly themselves, but often they don't -- seemingly on purpose. The *apparent* quality loss comes from a seemingly voluntary-added processing to consumer recordings, totally unneeded from a pure technical standpoint. Note: any comments made are 'observation' and not really 'creation', so there is no intent at causing discomfort to anyone...
Over the years, I have gotten a better and better 'bead' on the characteristics of the processing-in-common... There have been a lot of mistakes in my reverse engineering, but the results are starting to be on-par if not truly superior. Results are starting to be good enough to attract some interest from some moderately well known reviewers. (Not so-much for public review, but for personal collections. The project is still too questionable/hot-potato for those with 'skin in the game'.)
Very rough description of the processing technology:
The lower levels have a smooth sounding kind of compression, the audible effect is often just to 'raise the hiss' and sometimes to add a modulation fog that obscures subtle detail. The higher levels are usually handled 'okay', but there is still a slight amount of audible compression above 3kHz at all signal levels, not just low level... (It is actually a strange gain control pivot/teeter-totter around approx 3-4kHz, where the upper midrange has about 1dB of compression, and the super highs can have up to 6dB+ of compression, even at higher signal levels.) The level independent compression is so fast and has specific characteristics that it might be considered a 'strange' nearly distortion free 'soft clipping'. The non-level depedent phase basically bunches the spectral energy around specific frequencies, thereby softly rounding the time domain signal on some percentage of recordings. The 'spectral bunching' isn't all that effective, or at least my decoding attempts are not as effective as I want, but the experiments MIGHT be all that is possible. (The non-level dependent processing reminds of a known kind of processing similar to what is sometimes used in video.)
Not all recordings have this specific kind of both dynamics compression partially described as: below about -30dB, giving an amplification of hiss, plus some dynamics compression across the entire signal level range (near limiting) mostly above 3kHz, but some down to almost 1.5kHz. These impairments are ubiqutious, but not in all recordings, and there is the sometimes applied 'loudness wars' mess mostly starting in the middle/late 1990s, refined to a questionable 'art' in the 2000's.
My hypotheses is that this processing earlier described is a portion of the oft-complained (long ago) about 'digital sound' back in the middle 1980s when it seems to have appeared.
If you get a boutique recording or one of the select few 'really good' recordings, this description above is unlikely true. However, the results of the processing are especially obvious on old stuff like 'Brasil '66' or the Carpenters. The difference in hiss is even noticeable on some more recent, better respected recordings (1980s, sometimes beyond). Even on newer recordings, the level-independent dynamics processing in quite noticeable on, for example 'Call Me Maybe' from Carly Rae Jepsen (not my taste, just an observation based on trying the undo-processing on newer recordings, less likely to been 'processed'?.)
When trying to undo the processing on recordings that are not processed as described above, there is often a deadzone effect and subtle distortion, making the result unlistenable to a picky audiophile. Sometimes, the distortion is not so subtle. The result of trying to decode materials that were not encoded is certainly uglier than even the earlier attempts to decode the 'digital sound' add-on.
Semi off topic, no intention to 'shill' my project, but just example proof of the processing above (purposefully not giving pointers to it-- information available on request):
The 'decoding' project of the 'digital sound' is now producing reasonably good results, without the impairments of earlier versions caused by bad A/B choices resulting from my diminishing hearing. Luckily, the A/B choices are quantized and not continuous, so it is possible to eventually attain canonically correct results. (i.e. there are few real tweaks, and those tweaks needed are mostly forced to emulate legacy hardware. The actual tweaks in the decoding modules might be 1 or two out of 100's. The correct or relatively correct results are attainable because the settings are all based on standard dB increments and standard freqs modified to be related to a 221.5Hz basis frequency. Even with the dependency on my diminishing hearing, given the 3dB increments, eventually attaining canonically correct results is theoretically possible. The design appears to be architecturally complete, and the details are just wrapping up (painfully.) *The purpose of the project is for "existence proof" perhaps more so than for "decoding" the afflicted recordings for entertainment reasons.
From the article:
"[...] dividing it into two components: an audible low-frequency component (LFC) below 22 kHz [...]"
Audible, yes, below too. Still, apart from room modes not a lot of speakers can really reproduce this. Not in a home environment, I'd argue.
If this is different in your room, great. All the better, if you measured it, too.dividing it into two components: an audible low-frequency component (LFC) below 22 kHz
Edit: my bad, I misread and kept carrying this error with me (made "kHz" to "Hz" all the time, because my brain thought it had to be meant). My above argument about the bass is void.
EditEdit: I changed my first post above accordingly to point this out.
Actually analogue recordings (theoretically at least) are unconstrained when it comes to Frequency,. There's quite a lot of HF information on those old masters. Vinyl as a medium struggles with low frequencies rather than high. And super-tweeters have been around since the early 70s as i recall
I like songs, and groups of songs on albums
The Oohashi paper is not highly regarded by scientists as I understand it, or perhaps you mean a different one?
Last Century, Red Book being brickwalled at 22.05 was the cause & effect of the perception analog playback [LP specifically] had superior highs ['air'], even though an analog cartridge, phono stage, and/or the RIAA curve may top out at 20 kHz they're not brickwalled. Therefore it was surmised humans tended to perceive analog has a more 'open' sound.
Just personal perceptions that were shared back-in-the-day; posting because numbers don't always tell the whole story.
To be taken with a ton of sea salt!
The abuse of the CD format is an audiophile tragedy, I do agree.
Could you please name those scientists, as that paper has been frequently referenced by scientists at highly- regarded institutions such as Caltech.
Separate names with a comma.