24/96 and 24/192 is no good. MP3 sounds better!

Hi!

The subject is to get you going, sort of speaking.

Of course, it's just the opposite. I wonder though about the need for better and better techology.

http://www.diyaudio.com/forums/showthread.php?s=&postid=34048#post34048

The posting above had some reflections about advanced techology.

Why do we need 24/192 quality when almost none(!) CD recording is better than the 16/44 format. I wonder also is 192kS better than 96? Techincally it must be harder to sample at 192 than at 96 kS. Do you get sampling artifacts instead? Nothing gained really?

Some people don't want "spitzenklasse" for their amps but demands it for their other gear. We talk here very little about the speakers and I think in that area you have the largest possibilities to increase your sound quality. Not many people have well sounding speakers but the have good amps. Am I mistaken here?

I wonder also if anyone has heard a 24/96 and 24/192 recording which uses the full potential of the formats? I haven't but can't wait...very curious.
 
To say that MP3 is better is to strech it a bit I think. :)

I think it's true that the biggest faults still sits in the speakers and without good ones you won't benefit from super-DACs or amps.

There are quite a few recepies out there for making a great sounding amp, but not so many for making the typical great sounding speaker. It's of course due to the fact that everyone have different listening environments and different tastes in looks and sound.

I have cement walls in my listening room so nothing sounds too good there :mad:

Regards

Marcus
 
Personally, I can't wait for 24/192 to become more popular... at the moment there are only some 200-odd titles available, which is a bit dismal compared to SACD's 2000 or whatever it's at now. I for one am really looking forward to the improvement, since I do a considerable portion of my listening through headphones now... apartment living just doesn't seem to be so conducive to listening at the levels I enjoy (of course not really that loud, but enough to bother the neighbours). I also find it hard to sit down for long listening sessions as I used to do, thus it's a bit hard for me to justify warming up the ol' tubes...

Getting back to the issue, I've heard both SACD and 24/96 enough to realize that there's an appreciable difference compared to CD audio - especially for us audiophiles. Much to my dismay, however, it looks like SACD is going to dominate as the new high definition format. This is mainly because: 1. the Sony/Philips marketing machine is putting out good propoganda, 2. Sony/Philips own companies which produce the recordings, and therefore are releasing a good deal of material on SACD, and won't ever release anything on the competing DVD-A format, and 3. the SACD format is backward compatible with CDDA, thus making it easy for people to transition.

In contrast, there is little incentive for recording companies to release on the DVD-A format, since the current market penetration of DVD-A players is very small. On top of that, there doesn't seem to be much interest from the DVD-A companies to band together and push their technology against SACD. Perhaps the saving grace for DVD-A will be the addition of video / stills, a significant value-added feature for the typical consumer who might not be so concerned with the sound quality. I guess we'll just have to wait and see where the disc price point stabilizes, and how well the players catch on...

I'll be very disappointed if DVD-A loses out to SACD, as the SACD format is going to be pretty awful for us DIY-ers to deal with, especially if you want to process it. In order to manipulate the data in the digital domain, you pretty much have to convert it to PCM - negating the whole point of the format. But, at least 24/96 and 24/192 should be well capable of accurately representing the content of SACD.

In the meantime, I'm quite happy with my CDDA. The recordings and playback technology have come a long way, and it is still sometimes quite amazing what kind of fidelity can be captured on plain old 16/44.1. And, since you mention it, mp3s can sound OK too... clearly not up to audiophile standards, but I find higher bitrate mp3s to be quite listenable for background noise at home or work, in the car etc...

As an interesting sidebar, I recently discovered the following IC: NPC's new SM5816AF DSD-to-PCM converter. At least there's one relatively easy way to get the PCM bitstream out of DSD data... now, can we get our hands on these little devils?
 
SACD fan

The NPC SM5816AF seems like a really interesting part, although the datasheet seems to be a little sparse on details. But it appears to have some interesting features:

1) Writable coefficients
2) DSD pass-through
3) Multiple outputs 2fs and 8fs (fs = 44.1kHz, of course)
4) Flexible clocking

Does this mean that you can use this guy for a digital crossover? For example, you could use the DSD passthrough and pipe that directly to your tweeter amp (with an an analog bandpass filter) and take the PCM outputs and run that through digital room correction and additional crossover filters (there was a TI part that did this, I remember) for the <3kHz signals. You can refine this idea even more (for example, use the writeable coefficients creatively, or the extra set of PCM outputs), but I think this is pretty cool in itself.

I don't see any legal reason why we shouldn't be able to get this part because it doesn't seem like there's any sticky algorithm licensing issues here. Whether NPC would want to deal with sample quantities is a different story.

-Won

PS Ogg Vorbis is pretty cool, not just because it isn't proprietary (like .MP3) but because it is technically and subjectively superior.
 

Electro

Member
2002-04-12 4:41 am
The number one reason that 96 KHz or 192 KHz recording rates doesn't sound very good is because of data bandwidth limitations. For 16 bit at a sampling rate of 44100 Hz the designer needs a wide data bandwidth similer to RAMBUS. The ADC or DACs needs to be very, very, very fast to handle even CD quality sound. Analog computers is far superier than digital computers.

MP3 on the other hand is great for todays computers because it doesn't need a huge bandwidth. Most of the information in a CD quality sound file gets thrown away when recording in MP3 format.
 
Electro said:
The number one reason that 96 KHz or 192 KHz recording rates doesn't sound very good is because of data bandwidth limitations. For 16 bit at a sampling rate of 44100 Hz the designer needs a wide data bandwidth similer to RAMBUS. The ADC or DACs needs to be very, very, very fast to handle even CD quality sound. Analog computers is far superier than digital computers.

¿Que? What do you really mean?

My old slow Mac can in realtime process 16/48 and a couple of filters! A new Mac or PC can handle several channels of 24/192 in realtime and more if you have hardware acceleration. Man, Mac's can handle DV without problem! The bandwidth problem is really not a problem for playing gear. The biggiest problem is analog-digital conversion.

Maybe you aren't in the digital business but ADC and DAC chips are available to buy and they costs virtually nothing, we talk 24/192 and 24/96.
 

Electro

Member
2002-04-12 4:41 am
peranders, you don't understand. ADC and DAC a very slow. They loss to much bits. When converting sound to digital at 16 bits, it ranges from 5 bits to 13 bits. This is not true 16 bit sound.

Read paragraph "Latency Determines Resolution" in the document http://www.jam-tech.com/Class-J.pdf

My AMD 386 can handle sound but it takes a long time to manipulate CD quality sound. Though the speed is 40 MHz with out a math co-processor.

Maybe you aren't in the digital business but ADC and DAC chips are available to buy and they costs virtually nothing, we talk 24/192 and 24/96.
I already know that ADC or DAC are available to buy about several years ago. With school and about a hand full of hobbies that I do (programming, video editing, graphic editing, audio, computer building). It gets very busy.
 
Limits

The sampling rate of the original Redbook standard was based on the proposition from information theory that in order to exactly reproduce the original signal you needed a sampling rate twice that. Since human hearing maxes out (for female teenagers, everyone else is lower!) at 20kHz, 40kHz should do. This assumes perfect algorithms, of course, going some what higher would seem prudent although after 15 years or so of listenen to CDs, I think the 44.1k is nearly sufficient. Remember, ther IS an upper limit to audible resolution -- space between where we arew and "perfect" is rather small and set the upper limit on realizable improvement.

So I'll by the case for 96kHz sampling on the grounds of conservative over-engineering. 192kHz is just silly unless there is some other purpose such as carrying multichannel signals.

The word length may be a different matter. I don't know what the human limitation is on perception of gradations in amplitude. There are psychoaccouistic factors that I don't understand. Keeping in mind that adding one bit doubles the amplitude resolution, it would seem that since 16 bits sounds quite good, going beyond 24 bits (improving resolution 128:1) seems sufficient. This should allow icreases to the dynamic range of recorded music up to the maximum safe level.

However, there is another possible consideration. Without knowing much about specific algoritms used in music encoding but relying on many years around engineers and programmers, just having the extra bits and sample frequency available to play with ought to make digital signal processing products easier to accomplish. Code can be developed quicker if you don't have such tight constraints. Or to put it another way, you can get away with being quick and sloppy. That sounds bad, but for the consumer it's good because it keeps the cost down because engineers and p0rogrammers cost more than silicon!
 
I read a white paper by Bell labs that stated that the S/N ratio of human hearing, as limited by thermally induced noise in the auditory system, was approximately 120dB.

Having a full 22-24 bits can help, though, because the digital chain gets longer when we add signal processors etc and these contribute noise--this is related to your "sloppiness" factor ;).

As far as sampling rate, higher sampling rates mean that shallower filters may be used, and also increase the theoretical S/N per bit. There is also the issue of step response; we're now diverging into an area that is more opinion than theory, but it is plausible that a 192kHz DAC could more accurately reproduce transient attacks.
 
Transient attacks

I doubt any transient attacks from any musical instrument (save synthesizers) rise faster than the equivalent of a 20kHz sine wave and if they did I don't think humans could percieve it. Nonetheless, I'll grant there might be value signal processing advantages to having more samples available.

I suspect, higher sampling rate DACs will hit the market because of non-audio applications. They will get used in audio because they are cheap or have some other advantage in circuit design. The higher sample rate will be irrelavant except to the marking departments who will now have 384kHz DACs to help distinguish their product. Something similar goes on with high performance opamps. Thet operate at frequencies irrelavant to audio but have other advantages unrelated to the frequency.
 

Electro

Member
2002-04-12 4:41 am
I can't decide if you are a troll or really just clueless.
I'm not clueless. Many sites and books have experimented with ADC and DAC. ADC or DAC lose too much bits of information. The user can manipulate the audio in digital domain but it will always be different of what he or she intended.

...put it another way, you can get away with being quick and sloppy. That sounds bad, but for the consumer it's good because it keeps the cost down because engineers and programmers cost more than silicon!
Quick and sloppy drives in more complaints about buggy software and hardware. I rather pay extra for hardware and software that works 100% of the time instead of going back to the manufacture for a refund.


Sound is like drawing a picture on the computer. If the user decreases the depth (bits) that it was first created in. The picture loses quality. The user can always add more information or depth to the original picture although it won't change much.
 
sam9

Quick and sloppy doesn't need to mean buggy. Just the opposite sometimes. Anyone who can recall the wierd things done to try to squeeze code and data into a ca. 1970 mainframe will know what I mean. By sloppy I mean you aren't constrained to write the most succinct code imaginable even though it may be nearly impossble to comprehend the logic later or even if machine dependent quirks are utilized. These add to the likelyhood of bugs.

16 bits provides you with 65,536 discrete amplitude levels. I for one have never been able to hear a level change on a CD (including in sweep tones) that hint there any discontinuty is audible. 20 bits gives you 1,048,576 discrete levels and 24 bits gives you 16,777,216. Actually I don't think all the bits get used but just the same the progression is valid. Considering what is achieved with 16 bits, it hardly seems credible that within the relm of simple 2-channel sound reproduction that 24 bits can add anything of benefit to the listener. 20 bits maybe ... but even that a huge jump.

Another benefit I suspect but can prove is that the higher word size and sample size insulates the consumer (me!) from accumulated noise and distotion that adds up from all the steps from session to mix to master to pressing.
 
Electro,

***Warning Extremely Off Topic***

About Quantum Computing:

A few months ago, I visited NIST in Boulder Colorado. There I talked with some scientists working on quantum computing, with an actual quantum computer. They gave some sobering information. First of all, quantum computers are not being used by any government. They are still in their infancy. The most complicated calculation done was the factoring of 15. Yes, 3 and 5 are the factors. They are still working on the algorithims of quantum computing, as the data processing is nothing like standard computing. They *will* work best with doing many identical things at the same time. After they solve the scalability problems, factoring a 16 digit number will take as much time as factoring the number 16. This is why gov't's are interested in them. With Quantum Computers, one could decode *any* encryption, while making codes that only other quantum computers could crack.
What is interesting, though, is that quantum computers won't be fast at anything else until the algorithims become very clever.
After seeing the computer I must say this: don't hold your breath for a desktop version unless your desk is a 2000lb. air springed optical bench. :)

Sorry about the diversion folks. I'll try and keep it to a minimum from now on.

-Dan
 
i like SACD.

it sounds good. it doesn't sound like PCM. even 24/96 and 24/192 still sound like CD to me, albeit a lot better. but SACD has a naturalness of tone and timbre that no PCM i've heard can match... only good analog is really competitive.

i personally couldn't care less for DVD-A, and would gladly see it disappear from the face of the planet while SACD takes off (which it may not even if DVD-A dies). i'm biased though, given that a lot of the music i listen to is on Sony Classical which obviously gets released on SACD. PCM is easier to work with but SACD sounds better for my music, and i believe the A/D conversion is less complex. the D/A conversion can be a little tricky, with the noise shaping and all, and i don't think we've seen anywhere near the format's potential yet. PCM still has a ways to go too, particularly at higher speeds, but i think it's benefitted from a little more refinement over the years than DSD, which is newer technology... although i suppose single-bit delta/sigma converters have been around for a long time now.