creative new sound card tech (xifi) @ e3....anyone catch a glimpse/listen?

[tristancarton]

earlier this month creative announced a new sound card. this sound card (unlike the aud1, aud2,aud2zs,aud4) is not a slightly upgraded model of the emu/live chip. it is a brand new design.

saw (posted on their website) that they had a booth at e3, was wondering if any of you e3 kids got a chance to talk with them and see/listen to the product? (wondering price and release information as well as opinions)

[nrdstrm]

Yep..it was demoed for us (part of the Fatal1ty booth actually). Was pretty cool..Questions?

[jAkUp]

They did? Dammit :( I didn't see anything.

You saw it??!! I missed it then :( What was it like?

[tristancarton]

my main questions are the release date and price. also did they go into whether the card is capable of dolby digital encoding? encoding(like soundstorm), not decoding.
Yep..it was demoed for us (part of the Fatal1ty booth actually). Was pretty cool..Questions?

[nrdstrm]

Ummm...He didnt really mention that. The majority of what they showed off was making mp3's sound like CD/DVD quality, and making 2 channel headphones sound like 5 channel...
As far as release, I beleive he said sometime this fall...

[goaliechief]

Release date: august 21, 2005

[Rakeesh]

Ummm...He didnt really mention that. The majority of what they showed off was making mp3's sound like CD/DVD quality, and making 2 channel headphones sound like 5 channel...

You know, I hate when audio hardware/software companies do this. Basically they are telling you that you can get FM quality out of AM radio, and unfortunately, most people suddenly can't seem to tell the difference after they have fallen for the marketing hype and dumped all kinds of money into it. Sure they can make the AM sound better with a few tricks that may convince the average person that thinks deeper bass always means better audio, but it will never touch the fidelity of FM. And with regard to 2 channel being 5 channel, recall all of these 3d SRS variation sound technologies that claim to be able to, yet never do, replace a good ol HT sound system.

The worst part is that over the last 5 years, not a single solution for mp3 has ever come around that beats this (http://www.mars.org/home/rob/proj/mpeg/mad-plugin/) in terms of audio quality. Countless companies have tried, and insist left and right that their solution is best, but honestly none compare.

Creative worst of all. They have lied so many times about their products (and even got sued for lieing) that it's not even funny.

[killahsin]

You pretty much hit the nail on the head there. I think their 5 to 2 to 5 channel things is just positional amplification really. don't see any other way to actually do it. Should be cool to see. but i havent bought a creative card since the sblive card and neevr will again. Man that company is evil.

[jolle]

according to Creative it isnt simply upsampling to 24bit 96Khz, which would do nothing to help the quality, but its "filling in the blanks" as it upsamples..
Its said to sound good, havent heard it myself tho, and I dunno how of if it works..
But about 70% of the 10.000 MIPS they boast are dedicated to SRC.

[retsam]

Should be cool to see. but i havent bought a creative card since the sblive card and neevr will again. Man that company is evil. ya thats how ive felt .... im gonna wait and see.. i think this card might cause me to go back and buy one...but im also holding out promise that nvidia releases something along the lines of sound storm ....guess i just have to wait and see

[Rakeesh]

according to Creative it isnt simply upsampling to 24bit 96Khz, which would do nothing to help the quality, but its "filling in the blanks" as it upsamples..

This doesn't make any sense. That is effectively what upsampling does.

Think of it like this. Say you open a small image file in photoshop, and increase the image size to double the horizontal and vertical dimensions of what it was. Do you think it would make the image look better to just space the pixels apart one away from each other and leave the in between pixels white or black? Of course not. You either fill them in with the neighboring pixel, or you use some kind of filter (e.g. linear, bilinear, bicubic, etc.)

Audio is comparable, only in this case we are more or less one dimensional rather than two dimensional (or two dimensional rather than three dimensional when you take the bit depth into consideration, however you want to look at it.) Say you have an audio sample with 11,000 samples per second (picture this like the "resolution" of the audio, the bit depth doesn't really matter in this example,) and you want to upsample it to 22,000 samples per second. Do you leave every other sample blank? Or do you fill in the missing samples with their neighboring sample? Doing the prior will probably result in audio jitter, and somebody correct me if I am wrong here, but no common sound upsampling algorithms do this as far as I know.

The only thing I could picture them possibly doing is having some kind of audio equivalent of a graphical bilinear filter. Though I am not quite sure how this would benefit audio in any way. It benefits graphics by making the image look less blocky/pixely when you look at it up close. But then you can't really "look" at audio up close. You either hear it or you don't (though maybe you could play it in slow mo? not sure how this would benefit games or music.) The only thing I could venture to guess at them doing is maybe adding some kind of "speckle" effect (for lack of a better term) to make the audio sound different in order to mask the audio artefacting, but IMO all attempts at things like this make the audio sound worse.

[EciDemon]

Basicly what your describing is sound interpolation, and your absolutley right when comparing like you did.

I dont care much for "3d audio in 2d format" thing my self, it distorts the sound way too much, but sound interpolation can be very nice i think.

I can only speculate now but I dont think you would notice a big difference, but it could still be a nice thing ...

A good example of how sound interpolation is good is by looking back at the old 8bit sound era and the "mod" music format. Basicly all players for those formats has the optional sound interpolation and that really makes a difference. Unwanted scratchy noise is removed and in most cases it really sounds alot better.

So with a resonable good quallity mp3 song it could be nice.
but then again, the leap from 8bit to 16bit sound is huge and you notice it right away.
thats the difference of a 256 color image and 64k color image
16 to 24 is also a huge leap but not as obvious anymore.

However id rather see other hardware features like surround encoding, better asio support etc because thats something i use.

when I encode my own produced songs to mp3 for others to hear im pretty happy with the results

[Rakeesh]

A good example of how sound interpolation is good is by looking back at the old 8bit sound era and the "mod" music format. Basicly all players for those formats has the optional sound interpolation and that really makes a difference. Unwanted scratchy noise is removed and in most cases it really sounds alot better.

Hmm...wasn't mod basically a format like MIDI only audio samples of the instruments were included so that (at the time rather expensive) GM sound synthesis hardware wasn't necessary to get anything other than little beeps, hisses, or clicks for instrument sounds? The only reason I even know that format exists is because some of the later dos games started using it before it became common for soundcards to include their own GM synthesizers. :D I am not very familiar with the specifics of it though.

I forgot about sound interpolation though. I guess that could work, but would it really make a difference interpolating from 44.1khz to 192khz? You would be reconstructing very very small wave variations there, I would figure that interpolation wouldn't be able to do that with most audio (musical instruments would be ok, but something like e.g. weapons firing, car engines, voices, etc, are far more complex sounds.)

[EciDemon]

Yeah, mod format is very simalar to midi in terms of song structure, the major difference is that the mod format contains samples for the song too.

But it was only an example on where such a feature is good and used as we speak. :)

And I agree, the next step up is not as obvious so im not sure we hear such a great difference. It's a nice feature but really needed ?

[jolle]

I made a feeble attempt to illustrate what I assumed they meant.
http://upl.silentwhisper.net/uplfolders/upload9/kurva.gif
Right, the first curve is the "original" sound, analog or whatever.
Next the lack of precision in digital form.
Third a "pure" upsample, which should sound pretty much the same.
Fourth, what I assumed they were doing with Crystalizer when they upsample, sortof "fill in the blanks", that which should have been there, instead of just upsample straight off..
No idea about sound and that sort, just the impression I got reading about it.
No idea if thats whats being done, or how.

That is what you call sound interpolation perhaps?

A pure upsample shouldnt help, that would be like saving a severly compressed JPG in BMP and hoping for better quality.

[Rakeesh]

That is what you call sound interpolation perhaps?

A pure upsample shouldnt help, that would be like saving a severly compressed JPG in BMP and hoping for better quality.

What you are getting at looks like interpolation, but that example probably wouldn't work for the most part. Interpolation basically uses a sinusoidal regression based on the "points" of the digital wave to estimate how the originating analog wave pattern may have looked when it was in its original (pre-digital) form.

Now if your curved line was digitized as sharply as that one was, the interpolation will probably still reinterpret it as a sharpened wave pattern, at least if it has only those two points anyways.

For some low sample rate audio, this can help a bit on certain simpler sound patterns (e.g. from say upsampling a piano recording from 22khz to 44khz.) However, on more complex sound patterns (especially unmuffled car engines) this isn't going to help much. Especially when you go from 44khz to something higher. That is a bit like upsampling a 2400x1800 high resolution digital masterpeice (yes, some professional photographers actually use this) to 3200x2400. I suppose you *can* gain something out of it (though I am not sure what,) but if you really wanted that high of a resolution, you are much better off just capturing it that way in the first place.

EDIT: Actually this will explain it better than I can:

http://www.alpha-ii.com/Info/AudioInt.html

Specifically the examples I circled here where the wave pattern gets more complex:

http://www.nvnews.net/vbulletin/attachment.php?attachmentid=12985&stc=1

The $64 question is how much will this benefit games? Not sure what to say on that honestly. It could probably benefit music though (especially country, classical, or opera.)

[jolle]

The $64 question is how much will this benefit games? Not sure what to say on that honestly. It could probably benefit music though (especially country, classical, or opera.)
http://www.tomshardware.com/consumer/20050818/creative_x-fi-12.html
yeah it seems so, btw about 70% of the capabilities are dedicated to SRC.
The crystalizer works in all 3 modes, thus in games aswell, but they dont seem to mention much about its use in games, mainly music.

With many types of music, it'll be hard to go back to the way it sounded before once you've heard what it sounds like with the Crystalizer. Without it, you get the impression that you're listening to your music from behind a thick curtain.

[Rakeesh]

http://www.tomshardware.com/consumer/20050818/creative_x-fi-12.html
yeah it seems so, btw about 70% of the capabilities are dedicated to SRC.
The crystalizer works in all 3 modes, thus in games aswell, but they dont seem to mention much about its use in games, mainly music.

Yeah, the thing is though, all of these functions can be done in software, and in fact there are many programs available today that do this with many (all? more?) of the same features as the crystalizer with better quality sound as well since theres no up and down conversion necessary like the x-fi requires. If you are after this kind of functionality, IMO it would be more worth it to go for a cheaper software solution.

Games would be a different story, but we've yet to see if this actually helps games.