24/192 Music Downloads ...and why they make no sense

[rbro]

Interesting article.....

http://people.xiph.org/~xiphmont/demo/neil-young.html

[Tenshi]

I read this the other day and don't really understand why it exists. It's as if the person writing it lives in a parallel universe where the only available options are crappy mp3s or 24bit x 196k wavs. What I have heard is people saying "I don't want to pay 100% of the price for 10% of the music".

I mean, the article itself says something like the biggest factors include using a lossless format... lossless from what??

I also found it a bit grating that the article keeps telling me that I probably can't tell the difference and don't have the equipment to render it.

Crazy thought...maybe there is something between a 3meg mp3 and 300meg wav...

[c5_convertible]

I think you're not getting the point of the article.

It doesn't really discuss file formats (well, a small part of it does), but it's more a discussion of why 24/192 doesn't IMPROVE the sound (which is what a lot of people seem to believe) And the article goes on to say that even lossy formats, if encoded well, sound exactly the same as the original.

And it's also not so much about the equipment, but to your internal equipment, namely your ears and brain that CANNOT process more than what 16/44 can deliver. It says that the equipment CAN actually play it back, and that causes distortions in the part you can hear.

And there is something between a 3meg MP3 and a 300MB Wav... That's what the lossless compression is for. Lossless compression from the masters of course.

The article explains the idiocy of distributing 24/192 Files...

[Tenshi]

Who is distributing 24/196 files???

Last I knew beatport distributing cd quality wavs was unique exception.

And I don't get what you mean by saying it 'adds nothing' ... adds nothing to what? What is the baseline? As far as I can tell the baseline is somewhere between very crappy mp3s and good quality mp3s, in which case 24/196 files would DEFINITELY add A LOT.

I mean if we lived in a world where everyone was downloading 16/48 wavs as standard, and some company somewhere was planning to sell increased price files at 24/196, maybe this article would make sense. That's not the world we live in.

[c5_convertible]

This comes from the article:

Articles last month revealed that musician Neil Young and Apple's Steve Jobs discussed offering digital music downloads of 'uncompromised studio quality'. Much of the press and user commentary was particularly enthusiastic about the prospect of uncompressed 24 bit 192kHz downloads.

So, it is something that Apple was investigating... That's what the article was debating.

It adds nothing to the listening experience. There is no point in making a 24/192 file if you have a well encoded 16/44 file. There is no difference in what you will hear. Why take up more space for nothing? The article mentions as well that a 24-bit file also doesn't add anything as the dynamic range of a 16-bit file is more than enough. It does say that mastering with 24-bit allows for more headroom for badly recorded sessions (too loud or too soft).

A well encoded MP3 320Kbps file is indistinguishible from the original master, as the encoder takes into account what your brain (and ears) can hear. They are NOT the same as the original master though.
In double blind testing the correct file (good MP3 vs original) could be identified with a little less than 50% accuracy. Which is the same as you would get when flipping a coin...

[Tenshi]

You have to realise discussions in Apple between high ranking people probably refer to things that will happen over 10 years, not next week.

If we fast forward 10 years downloading a 300 meg file will take no time. I can see an overall solution where people download a single master version of a track then that is coded into whatever format is best for the devices they sync it to.

Also, there is already a movement to market high quality headphones. Beats by dre are already selling studio monitor level headphones as consumer headphones (at 150% the price).

Also the thing about sound 'quality' is an open debate. For example, perception of 3D position requires timings of less than the 1 sample at 44.1kHz. Its not just about pitch.

Funny story: the other day I was behind someone in a queue who had a pair of senheiser monitor headphones... listening to an ipod.

[c5_convertible]

I doubt that anyone can hear this... 1/44000 of a second is 0.0000227 seconds. Or, 0.0227 ms.
A time delay of 0.1ms can be heard (according to many studies), this is 5 times more than one sample. 3D imaging is the difference in time between the left ear and the right ear. To differentiate, the brain needs 0.5ms if it is in the left to right plane. In the up-down direction it needs even more time delay.

Also, it doesn't matter than 10 years later on we can download the 300MB file at the same speed as now a 3MB file, as the added size does not add anything in what we can hear. Why download 300MB (even at lightning speeds - I can download this in about 45seconds now as well), if the same file of a smaller size results in the same quality?

But I agree with you in as far as it allows options. But, again, that was not what the article was about.
And it is indeed an open debate.

as a side note.. I don't consider Beats by Dr Dre good headphones. They are too bass heavy. Not good for mastering. But that is my opinion of course... The ipod's output is actually not that bad. The earphones that come with it are worthless. I use Koss PortaPros for my iPod, which is a lot better. All my MP3's are encoding in 320Kbps, and about 6MB in size

[Tenshi]

On the topic of the brain, I happen to be in that research area and have read a lot about how it works and can tell you that talks about its 'limitations' are often wrong.

I doubt that anyone can hear this... 1/44000 of a second is 0.0000227 seconds. Or, 0.0227 ms.

I didnt say you can 'hear' it. That doesn't mean it doesn't form part of the way your brain constructs your perception of what you hear.

Inter-aural delay is 660?s at maximum and 0 at minimum, but we can break direction down to something like 1 degree increments. That means there are 90 increments between forward and left. Thats 7.3?s between direction increments. Knock a sound out by 7.3?s and we can hear the difference.

One of the most startling things I remember learning about the brain is that you can construct a visual perception of a scene in a time less than the time it takes a neuron to fire, which is conceptually nonsensical.


BTW: Ofcourse, I'm not saying the are good mastering gear. But they seem to be selling something at that range.

[Tenshi]

Also it occurred to me that this relates to the whole loudness war thing.

Image we could download a floating point version of the track and choose the level of dynamics we prefer using post processing at the listeners end.

[NothingButInitials]

Interesting article for sure... The part I don't quite understand is this:

"I've run across a few articles and blog posts that declare the virtues of 24 bit or 96/192kHz by comparing a CD to an audio DVD (or SACD) of the 'same' recording. This comparison is invalid; the masters are usually different."

"The BAS test I linked earlier mentions as an aside that the SACD version of a recording often does sound substantially better than the CD release. It's not because of increased sample rate or depth but because of the better mastering of the SACD. When bounced to a CD-R, the SACD version still sounds as good as the original SACD and much better than the CD release."

The author doesn't give any explanation as to why the DVD-Audio or SACD versions would be mastered differently than the CD version. Why master it twice if "When bounced to a CD-R, the SACD version still sounds as good as the original SACD and much better than the CD release."

Is there perhaps a reason besides needing to have a "better" sounding DVD-A or SACD version in order to justify the higher price?

[Goddard]

Comparing the sampling rate to the audible frequency?
To state that it's pointless to sample with rate of 192kHz, because it's above the audible range you have to be a complete moron.
I rest my case. These kind of bollocks fits in this forum well...
Greetings.

[c5_convertible]

Tenshi: Thanks for the explanation. I learned something new here
Indeed, the whole picture (or in this case, sound) is constructed in your brain. In that I follow you. What I believe the article is trying to say is that at some point, there is no benefit in going to higher sampling rates and or bit depth, except for using more disk space and resources. Also, I believe that what you talk about (7.3µs) is when there is only one sound going. In a normal (real) music piece, there are several sounds layered, which will automatically decrease the differentiation you can make, and hence, will also influence your perception.

Goddard: Did you just call me a moron? What is the point in sampling at 192Khz then? So you can sample something only bats can hear?? A sample rate of 192Khz results in samples of 5µs, which is below the threshold that Tenshi talks about.
At the same time, your sample takes up 4.36 times more disk space (or memory).

[Tenshi]

(7.3µs) is when there is only one sound going. In a normal (real) music piece, there are several sounds layered,

Sure, but there is never really such a thing as 'one sound' versus 'many sounds', you ultimately have 2 ears and therefor 2 2 dimensional signals. If you hear a sound in real life you hear it plus lots of other sounds, and they all get mixed down into the same 2 channels... then split into god know how many other channels in your neural network.

[trevox]

Tenshi: Thanks for the explanation. I learned something new here
Indeed, the whole picture (or in this case, sound) is constructed in your brain. In that I follow you. What I believe the article is trying to say is that at some point, there is no benefit in going to higher sampling rates and or bit depth, except for using more disk space and resources. Also, I believe that what you talk about (7.3µs) is when there is only one sound going. In a normal (real) music piece, there are several sounds layered, which will automatically decrease the differentiation you can make, and hence, will also influence your perception.

Goddard: Did you just call me a moron? What is the point in sampling at 192Khz then? So you can sample something only bats can hear?? A sample rate of 192Khz results in samples of 5µs, which is below the threshold that Tenshi talks about.
At the same time, your sample takes up 4.36 times more disk space (or memory).

I think the title of thread makes a bit of sense as it speaks in terms of listening to completed pieces of music, and I think listening to music at 24/192kHz is overkill.

However, there is a big difference between listening to music and creating it. Sampling and processing audio has nothing to do with downloading completed music and is not reliant on our ability/inability to differentiate between something recorded at 16/44.1kHz or 24/192kHz. For instance, if you start with a 16/44.1kHz file and apply say 4 or 5 effects and do the same with a 24/192kHz file, you may audibly hear the difference in the resulting audio - particularly if you are using convolution reverbs and the likes. And that's not to say one sounds better than the other either - just that the results may be audibly different.

Talking solely about sample rate, a simple example would be this - if you sample something at say 48kHz and pitch it down 2 octaves, you essentially end up with a sample 4 times the length with (in effect) the same number of samples and it will sound grainy (it would be the equivalent of a file recorded at 12kHz). Do the same to a file recorded at 192kHz and you'll find you have a sample that does not sound grainy (it would be the equivalent of a file recorded at 48kHz).

I know I have simplified things a bit here and there are other factors, but I just wanted to point out that there are uses for sampling at 192kHz as while we may not be able to distinguish between 2 audio files, the computers and other audio processing devices we use can. So until the final file to be listened to is created - i.e. a released track - there are reasons to work using higher bit/sample rates at times.