Is 48Khz "sufficient" for audio?

[SaltBoy]

I've noticed that some places that sell flac files for download at 24bit are offering them at 192Khz. Is it just me, or isn't that incredibly stupid?

From what I understand, the frequency of a file just tells how high the "pitch" could be presented as the file plays, right? CDs are at 44.1Khz, which makes sense but many DTS-HD or Dolby True HD soundtracks are at 96Khz; that doesn't necessarily make sense to me but I can live with it considering the fact that they represent a copy of the studio master.

Wouldn't playing a 48Khz file be sufficient in the "real world"? Or maybe, even a lower frequency could work since the human hearing range is normally around 20Khz?

Of course, more bits is better, and yes, I understand the draw with switching to Vinyl entirely (scratchy noises notwithstanding). But in the end, wouldn't buying something that advertises 192Khz just a useless waste of space? Heck, would somebody be even dumb enough to include a pitch above 48Khz in their music? Do they intend on providing subliminal messages to my neighbor's dog...?

Of course, I just might not know what I'm talking about at all... If somebody could educate me, I'd appreciate it. 🙂

 

[sdifox]

192 is nice. If your system can play it back well. And by system I include the room.

and DTS-HD and DTHD are not representations of studio masters. far from it.


Have you actually listened to DVD-A or SACD?



edited out my stupid stupid comment

 

[PhoenixEnigma]

192 is nice. If your system can play it back well. And by system I include the room.

and DTS-HD and DTHD are not representations of studio masters. far from it.


Have you actually listened to DVD-A or SACD?

You were right, you really don't know what you are talking about.

Are you thinking 192kbps? That certainly makes a difference over lower bitrates - not one that every person will be able to hear or care about, but certainly one that can be noticed under the right conditions.

A 192kHz sampling rate with any kind of remotely decent sampling is way, way overkill for reproducing anything the human ear can hear. Generally speaking, you need a sampling rate twice that of the highest frequency you want to reproduce. Human hearing is often quoted as topping out at 20kHz, though if you do a frequency sweep downwards most people seem not to hear anything above about 17kHz. 44kHz and 48kHz are both sufficient to reproduce that without problems - arguably, 48kHz is a little safer, as people who can hear 22kHz aren't exceedingly rare, but 24kHz hearers are.

Recording at higher sampling rates can be useful to prevent artifacting when manipulating the sound, but there's no reason for it during playback.

 

[sdifox]

Are you thinking 192kbps? That certainly makes a difference over lower bitrates - not one that every person will be able to hear or care about, but certainly one that can be noticed under the right conditions.

A 192kHz sampling rate with any kind of remotely decent sampling is way, way overkill for reproducing anything the human ear can hear. Generally speaking, you need a sampling rate twice that of the highest frequency you want to reproduce. Human hearing is often quoted as topping out at 20kHz, though if you do a frequency sweep downwards most people seem not to hear anything above about 17kHz. 44kHz and 48kHz are both sufficient to reproduce that without problems - arguably, 48kHz is a little safer, as people who can hear 22kHz aren't exceedingly rare, but 24kHz hearers are.

Recording at higher sampling rates can be useful to prevent artifacting when manipulating the sound, but there's no reason for it during playback.

Nope, talking about sampling rate. DVD-Audio format allows 192kHz 2channel 24bits. I should really listen to my dvd-a more... HD music died a horrible death because most people are content with lossy mp3 they pirated

edit:wtf is glossy mp3...

 

[Belegost]

48KHz is quite good for human audition purposes. In fact, oversampling beyond that only adds possible distortion caused by imperfect amplification giving rise to lower frequency (actually audible) artifacts.

As for 24bits... very important if you are mixing several 16 bit sources to ensure you capture the full dynamic range of all sources without going through the trouble of registering the quantization scales. For playback... 16bit quantization noise out of a half decent quantizer is well below a reasonable noise floor.

This is a good source for a quick overview and some references
http://people.xiph.org/~xiphmont/demo/neil-young.html

 

[sdifox]

48KHz is quite good for human audition purposes. In fact, oversampling beyond that only adds possible distortion caused by imperfect amplification giving rise to lower frequency (actually audible) artifacts.

As for 24bits... very important if you are mixing several 16 bit sources to ensure you capture the full dynamic range of all sources without going through the trouble of registering the quantization scales. For playback... 16bit quantization noise out of a half decent quantizer is well below a reasonable noise floor.

This is a good source for a quick overview and some references
http://people.xiph.org/~xiphmont/demo/neil-young.html

Interesting article. I am no sound engineer, but I do like my DVD-As.


I guess I owe the op an apology.

 

[Howard]

Interesting article. I am no sound engineer, but I do like my DVD-As.


I guess I owe the op an apology.

It feels like a biased article. I'll look over when I get home but there's definitely a slant there; it's the science that the author wants you to see.

 

[sdifox]

It feels like a biased article. I'll look over when I get home but there's definitely a slant there; it's the science that the author wants you to see.

I always understood that perfect representation is not possible since we cannot sample infinitely, which is the base assumption of Nyquist. Aliasing happens when you are not sampling infinitely.

My response was still childish though.


Who's got access to IEEE archive? would love to look at their papers on MLP

 

[Rio Rebel]

... HD music died a horrible death because most people are content with glossy mp3 they pirated

You and I are SO in agreement on this. I consistently become more and more saddened at the way the public favors audio convenience over quality. You can't hate them for their taste, but it is sad that we can't utilize the technology available to get the best of all worlds. Audio COULD be incredibly high fidelity, convenient, and affordable...but since convenient and affordable are the only drivers for 99% of the public, quality is simply ignored as long as it's not crackling and popping.

 

[sdifox]

You and I are SO in agreement on this. I consistently become more and more saddened at the way the public favors audio convenience over quality. You can't hate them for their taste, but it is sad that we can't utilize the technology available to get the best of all worlds. Audio COULD be incredibly high fidelity, convenient, and affordable...but since convenient and affordable are the only drivers for 99% of the public, quality is simply ignored as long as it's not crackling and popping.

But the rich will pay extra for the crackle and pop :whiste:

 

[ImpulsE69]

Remember that people grew/grow up on radio and TV. Radio quality is not high at all. Anything better than that they consider "good".

There is no reason we can't have HD quality MP3's.

 

[Rio Rebel]

Look at blu-ray. The opportunity is there to have a whole new level of audio quality for recorded concerts, but the amount of bd content for music is still sparse. I would spend all sorts of money to get my favorite music on bd in TrueHD or DTS MA.

 

[Anarchist420]

48 KHz can be enough... I'd be more than satisfied if everything was 16 bit 44.1KHz and not lossy at any stage, but that's not the only issue... some people even like a ****load of harmonic distortion. Also, since we are dealing with digital signals and binary code (bit rate not just frequencies), it is not just about the frequencies the "human ear" (which is a very general term) can hear. The fact that digital processing can be overly used makes some masters sound better than others subjectively.

 

[Bryf50]

From what I understand, the frequency of a file just tells how high the "pitch" could be presented as the file plays, right?

This is incorrect. The frequency of a file is the sample rate. When you are converting an analog signal to digital you take a sample at certain intervals(the sampling rate). The reason why sampling rates in the ~40khz range are most common is because according to the sampling theorem you need to sample at least more than twice the frequency of the range you wish to reproduce(~20khz for human hearing).