Is there a technical limit on the sound quality of PC music formats?

But they are both large file formats. You are very limited storage wise if you choose to use these formats exclusively.

Unless storage prices drop infinitely, I can't see lossless digital formats taking over.

Who wants to start archiving lossless files? It defeats the benefits of digital.

Which few? News to me (not that I believe it).

(Shitty 128kbps ones however are common and were considered "acceptable" for a while, and have a quite easily recognisable sonic artifacts (smushiness, distortion) that mark them out almost as clearly as tape vs. CD vs. vinyl.

Also, I woke up really f*cking early this morning and the entire topic is stupid and embarrasing now. Of course there isn't a 'limit', I don't think so anyway.

I have around 200 gigs of flac files. Archived on DVDs. The lossless trading community does this kind of stuff all the time.

I thought this question was going to be about 24/96 sound cards or something.

That's probably where my fuzzy thinking was headed this morning yes...

the price of hard-crives these days make it quite reasonable.

24/96 probably wont' take off ever.

Interesting article on music being remastered to hide the deficiencies of file compression:

Music throughout the last decade is typically recorded using 24-bit samples at 96kHz, and advances in computing power and hard disk space have recently made even higher quality, 24-bit 192kHz digital recording possible.

However, even the standard CD format comes in a much lower resolution—just 16-bit 44.1kHz. Compared to 24-bit 192kHz digital audio, a finished CD only has roughly 15 percent of the information captured during the recording process. Compressing the songs on a CD further into 256kbps AAC "iTunes Plus" format cuts the data down to just one-fifth of the size of CD audio, or as little as three percent of the original 192kHz recordings.

http://arstechnica.com/apple/news/2012/02/mastered-for-itunes-how-audio-engineers-tweak-tunes-for-the-ipod-age.ars

Except that, as one of the commenters on that article pointed out, CD reproduces 100% of the audio spectrum that can actually be heard by an adult human being.
Also, 24-bit, 192kHz recording has been widely available for years.

yeah but then it's difficult to tell to what extent that which we can't hear is still affecting us in some way, isn't it? obvious example being super-low sub-bass and lower.

The resolution of CDs at the high end is pretty rubbish.

^ I think you're half-wrong, really. I can't tell the difference between CD quality and 16 bit 96 kHz. (24 bit 96 kHz I can detect the *smallest* difference but only on cymbal decay.)

The rubbish part is in the dithering process. Still in its infancy. My preference for vinyl, hopefully pressed from analog sources, or 24 bit 96 kHz-- my preference for vinyl is simply because I can detect dithering, and I think as dithering gets better over time, it'll make our current transfers sound lousy.

The resolution of CDs at the high end is pretty rubbish.

Citation? I mean, I don't agree, so I'd like to know your evidence for that assertion.

I think as dithering gets better over time, it'll make our current transfers sound lousy.

Hmmm... dithering is the application of low-level white noise - sometimes spectrally-shaped - to a digital audio signal, so as to ameliorate the effect of quantization errors. This is the first time that I've heard it suggested that it's a magic bullet?

I said "dithering" but I meant, specifically, the conversion of higher quality audio formats down to lower quality formats, which has been, in my experience, referred to in the mixing and mastering process as dithering, although actual "dithering" may or may not be involved.

Personally the "magic bullet" with regards to the development of digital audio is still in the world of digital summing, and analog-digital/digital-analog conversion.

assuming there's nothing digital in the recording chain, what's the equivalent bit-depth for vinyl?

It doesn't apply, there's no digital conversion so there are no bits.

well phrase it another way, how much information could you pack and unpack on any given slice of vinyl.

Ówen P, POW-r #1,#2 or #3?

Citation? I mean, I don't agree, so I'd like to know your evidence for that assertion.

Just that 44.1khz means you're only sampling twice per wavelength at the v high end, so you must be losing a lot of detail, right? Idk if it makes any audible difference.

xp Infinite information. You could theoretically put a terrabyte of info on a 7", provided that you had some manner of encoding/decoding it.

@ owenf Is that an Ableton thing?

well it's settled then, we need to develop an infinite digital file format. FLAAAAAAAAAC or something.

Digitally, 96 kHz is good enough to work with, 44.1 is good enough to listen to. I'm much more interested in Ambisonics, when it comes to encoding and decoding. That stuff is fascinating!

haha no it's an across board dither noise shaping algorithm. Think it's offered in ableton though (although probably called something different) , as well as almost all DAWs and a lot of limiting/dither plugins. I find it interesting that it's offered to casual musicians recording as well as mix engineers (who IMO shouldn't be allowed to do that sort of thing) at the bounce stage

all three that is, sometimes called light, medium and weighted etc and apogee converters have their own algorithm too I think

*eyes bugging out*

what's the equivalent bit-depth for vinyl?

A commenter on the AT thread states that "the constraints of vinyl are about 12 bits dynamic range". That's an estimate that I've seen before and it amounts to a significantly lower fidelity than is available from CD.

24-bit samples at 96kHz

pretty awesomely you can download some recordings from true nerd artists at this bit depth/frequency. new monolake album is supposed to have such a release.

xp to VastHale re: dynamic range

I suggest checking the Vinyl Anachronist archives on the Perfect Sound Forever site because I think that is almost certainly untrue, in my reading and accumulated knowledge I have consistently heard that high-end vinyl really clobbers CDs in terms of dynamic range.

I want to say that FLAC proponents were saying that Apple Lossless (ALAC) lacked something or another, but what I just read seems to imply that the format supports 24 bit / 96kHz as well. Maybe some of the current portable devices don't work correctly with it? In any case, there's no real reason that as bandwidth increases that Apple or other services couldn't just start distributing lossless. With movie downloads being more common right now, I don't know why they wouldn't.

The next thing beyond that is multi-channel and whether formats support it and whether anyone really needs it.

I'd just be happy with lossless! It's more common on Bandcamp these days.

I was reading something about soundcloud today and the lossy encoding it uses. Was a link on from a thing on engineer types mastering for particular mp3 formats that were becoming popular. For example if something was going to be a free download via whatever hosting site, they would find out how the hosts are encoding uploads and making mid/high frequency adjustments accordingly at the mastering stage.

will grab that link in the morning

Dynamic range in terms of volume shift from quietest to loudest is about 75db for vinyl and 90db for cd, was my understanding. Unless you guys mean some other kind of dynamic range.

tbh I think it was getting misused as frequency? Not sure

the width of the grooves is the limiting factor for vinyl.

as an analogue recording device there is no actual limit in the way there si for digital. as you are capturing the waveform not a sampled facsimile of it.

but the louder and the more dynamic range you get on vinyl, the harder it is to keep the needle in the groove, or not to have a platter that wears out immeadiately.

Just that 44.1khz means you're only sampling twice per wavelength at the v high end, so you must be losing a lot of detail, right? Idk if it makes any audible difference.

― ledge, Friday, February 24, 2012 6:42 PM (Yesterday) Bookmark Flag Post Permalink

Sounds approaching 22kHz are difficult to distinguish tonally anyway; they're felt more than heard as a clearly defined tone. afaik the only reason things are often recorded higher than 44.1kHz (eg 96kHz) is to avoid aliasing.

Yeah, 44.1 khz was chosen since 22.5 khz was seen as the upper end of human hearing (and most adults are well below that). But you still get problems with things like cymbals and trumpets, which have harmonics going on towards infinity. The filtering of those tones has to start well below 22.5 khz, so as to be brought in gradually before hitting its limit; if poorly done, you get rippling and aliasing. Which is one of the reasons recording gets done up at 88.2, 96, or 192 khz--it keeps the problems well above where the human ear can hear them. But you still have to have good filters in place each time you step down/convert.

there is no actual limit in the way there si for digital. as you are capturing the waveform not a sampled facsimile of it

Hang on a minute - vinyl is a medium that has definable limitations, just as CD does. The groove on an LP is an analogue representation of a waveform, but that does not mean that its resolution, and hence its fidelity to the source, is infinite. Cassette tape was also an analogue medium, but no one has ever suggested that cassettes rivalled CD for audio quality.
Vinyl always represents a compromise. For example, the first thing a mastering engineer will typically do when cutting an acetate is to apply a high-pass filter, to roll off the deepest bass frequencies in the music. Those frequencies pose no problem to a digital medium, but could cause needles to hop right out of the groove.

Just that 44.1khz means you're only sampling twice per wavelength at the v high end, so you must be losing a lot of detail, right? Idk if it makes any audible difference.

Nyquist says that you only need to sample at twice per wavelength of the highest frequency to get a perfect copy of what went in. It's not like pixels in digital images.

I was reading something about soundcloud today and the lossy encoding it uses. Was a link on from a thing on engineer types mastering for particular mp3 formats that were becoming popular. For example if something was going to be a free download via whatever hosting site, they would find out how the hosts are encoding uploads and making mid/high frequency adjustments accordingly at the mastering stage.

I would like a link for this because anything I've mixed myself and uploaded to Soundcloud sounds like complete wank.

Twice per wavelength to satisfy Nyquist, but keep in mind the filtering to cut off those harmonics (brass, cymbals) that extend beyond 22.5 khz actually has to start considerably before that. To have enough room at the top to support a gentle filter curve (low pass--letting every thing lower go through), it's a good idea (and I'm basically quoting Bob Katz here) to have at least a 50khz sampling rate, or something significantly beyond CD's 44.1., and a little beyond DAT's 48 khz. And note that the problems are not coming so much from the original audio (whether at 44.1 or something higher as from poorly implemented filters.

As for mastering for different codecs, I've never done it, so take this with a grain of salt, but I think more and more engineers are running their work through plug-ins like this one in order to test the results with different coding schemes, perhaps even to the point of doing different masters for different codecs:

http://www.sonnoxplugins.com/pub/plugins/products/pro-codec.htm

"Hang on a minute - vinyl is a medium that has definable limitations, just as CD does."
the CD spec is pretty well defined.
the bit-depth of vinyl is apparently somewhere between 12 and infinity.
re: cassettes, i suspect better-than-CD fidelity is possible provided recording/playback machines are sensitive enough, and there might very well be some tapes that were prepared to that extent -- are there any famous examples?

the bit-depth of vinyl is apparently somewhere between 12 and infinity.

This doesn't make any sense. The problem is you cannot make a direct comparison between digital and analogue in that way.

One of the measurable things with digital recording at different bit depths is noise levels -- a 16 bit recording has much lower noise than a 12 bit one. Of course that is not the only difference. In a sense, vinyl might be comparable with 12 bit digital **in terms of its noise levels**, but in other respects the quality is potentially much better. On the other hand there are problems with vinyl such as wow and flutter, plus wear and tear on the disc after repeated playback etc. etc. Digital does not have those problems but it has problems of its own, of course.

re: cassettes, i suspect better-than-CD fidelity is possible provided recording/playback machines are sensitive enough

Better in what sense? The noise performance of cassette is inferior to that of vinyl, which is inferior to CD. You can use Dolby noise reduction to reduce noise but otherwise you are talking about quite obvious hiss/noise that would be absent on CD (assuming such hiss/noise is absent from the original recording from which the CD is made). Cassette also has the same wow/flutter problems as vinyl. I know there is a huge difference between a low grade cassette recorder and a top of the range model, but even at the top of the range I don't think you can reach the noise performance of 16 bit digital. Of course, that's not to say that good quality analogue (whether on vinyl or cassette) might in some senses sound as good or better than CD, because of considerations other than noise performance. (I've heard some people say of tape hiss that they are able to disregard it because in a sense it is perceptible as separate to the actual programme material. I kind of agree with that, although typical noise levels on cassette are pushing it in my opinion.)

"The problem is you cannot make a direct comparison between digital and analogue in that way."
I agree that it's not easy, but maybe a new digital format could incorporate variable bit rates or some new scheme for audio representation to better reproduce analog.

"Better in what sense?"
better in that a tape dub could feasibly capture more information than a CD transfer, just because the limitations of how much extractable information is recorded on a CD is set by the spec, rather than the equipment and materials. It would be interesting to see where tape technology would have progressed -- would we see cassettes with the same performance as master tapes now?

apparently EVERYTHING soundcloud streams is at 128 which explains why it makes stuff sound shitty however as a download you are downloading a file that will have the original bit depth etc as you uploaded it. Can't find the exact link but I think it stemmed from a gearslutz thread.