This little white box can do magic, apparently

This is the Creative XMod, a little €80 box which plugs into your computer as a soundcard. Their main claim is: "X-Fi module for PC and Mac restores MP3 music beyond studio quality". It's not clear how they perform this wonder, but they say: "The Creative X-Fi Xtreme Fidelity audio processor puts your music and movie audio through a two-step quality enhancement process. First it converts the audio into 24-bit/96kHz quality, then it remasters and selectively enhances the audio by analyzing and identifying which parts of the audio stream have been restricted/damaged during the compression stages to 16-bit and then to MP3. The result is music that sounds cleaner, smoother and has more sparkle, and movies that sound more realistic than ever before!" Could it be a touch of EQ and a compressor, perhaps? Whatever, let's hope that some professional studios buy one quick, so they can catch up, soundwise. (Thanks, Taras)

so it knows what to add on the top and the bottom?

it better be mac compatible.
maybe a "creative quality" exciter?
80 dollars??? i'll search for an aphex
When it converts to 24-bit/96kHz, it has some "intelligent algorithm" for interpolating the values between 44.1 kHz samples. So you are getting smooth waveforms when you upsample, instead of it being squared off, and the interpolated values are shaped (so it isn't just a linear change or simple curve).

Also, when increasing the dynamic range from 16 bit to 24 bit, it figures out a smarter way to set the dynamics that just taking the closest 24 bit value corresponding to the 16 bit value. It looks at the change of dynamics, tries to look at the curve of dynamic change, and make an intelligent guess where it should be.

Of course, the device itself is stupid, as you could just create a small sofware application which would do the same thing to your file, and output a 24 bit 96khz wav file or something.

Or, there is the fact that will consumer electronics, there is no noticable difference between 16bit 44.1 khz and 24bit 96khz (if you claim you can hear the difference, you are probably using expensive studio monitors or audiophile equipment... you aren't listening using ipod earbuds or a cheap Sony bookshelf stereo. NO-ONE can hear the difference on ipod earbuds or a cheap stereo, no matter how good their ear is!)
Except that converting to 24-bit can't get back the information lost to quantization, and converting to 96KHz can't get back the lost phase information, wich if you're not going to be doing any further editing is really the only useful thing a highers sampling rate will get you.
nobody has said it yet...

better than studio quality!

(what does that mean really?)
it's a must have for All mastering studio!
For 80$ it's supposed to increase the master quality from 10% (according to the diag'). it must be a kind of anti-dithering tool that interpolate the bits scale to the upper quality. what a dream :)

I will ebay all my favorite gears and my crappy vintage compressors and enhancers to buy a soundblaster and this magical apple-design-ripped knob.
let me laugh.
I'm really looking forward to hearing this crap box with a Hi def master tape...

Viva Creative!
What a joke. I saw this the other week and thought WTF?!
gotta love marketing goons
let's all paypal a euro to TOM and have this thing checked out!
oh yeah, this is all the rage nowadays.
you know when you go "Oh shit! I meant to record a harmony there" or "Shit, that drum hit should be a snare!". This little box does it all for you.
Better than studio quality. We love it.
can it make a 64 kbps mp3 sound like studio quality? then i'd be impressed.
ha ha! Top notch entertainment. A fool and his money are soon parted. It's rather foolproof in a way though, because if you don't know the difference between lossy and lossless then you'll doubtless know that this works 100% right?
If you're still not convinced, pump it through your 1" multimedia speakers with 3D surround sound, everyone knows they replaced reference monitors ages ago.
...for people who record, produce and mix their music entirely on a windows PC with a soundblaster card, a 64kbps mp3 is about as good as it gets anyway so this magic gizmo can't do any harm eh.

Turned out nice again ain't it?
This reminds me of the compression crackpots who turn up occasionally (sometimes with a "business") claiming to be able to compress an arbitrary number of bits to one bit (lossless, obviously) or some such thing. This is a compression issue. Compression is reasonably well understood.

By the way, somebody up there seems to have "dynamic range" mixed up with "resolution".

Maybe they've got an algorithm that breaks up the whoosing/underwater sound of low-bitrate mp3s so they're slightly less irritating? But it's probably not even that. It probably just amplifies the signal a little bit, and maybe boosts the bass and treble. A/B a signal like that with the original and 99.9% of listeners will be stunned by the "improvement".

If it were some kind of interpolation deal, where would that get them with aliasing? The information is gone.
im not sure the information is totally gone.. as you can decode an MP3 file back to wav format to regain a bit of the original quality - its not the same as the original file tho

maybe this x-fi garbage has something to do with the realtime output processing

altho - how good can it really be? especially when it claims to be better than the recording studio that produced the fucking song in the first place? come on with that shit
There have been numerous algorithms that do something like this, including a number of software plug-ins for Windows Media Player and whatnot. This is just another. There are ways of interpolating missing data, so that's possible, and actually you would want 24-bit/96kHz just to give you headroom for the conversion. But having heard the results, generally what happens is they just "soup up" the content. Basically, if your audio is over-compressed enough to need this, it'd probably too far gone to sound good anyway. (...and most well-compressed audio, while not perfect, is listenable on its own)

The algorithms do *something*, and they may make things sound better, but this seems like a weird time to reduce this gadget as compression has gotten better, bitrates are up, and lossless is available.

And, of course, all this marketing stuff is completely misleading.

Sooo.... on the x-axis we have audio format. Fair enough. .... But, on the y-axis we have experience?!
Can someone please tell me how they are measuring experience? I mean, what units could they possibly be referring to? Maybe smiley faces? Or better yet, aural orgasms. Lol.

This little thing is brilliant! :)
I figured it accessed the global conscience, extrapolated the lost quantized parts by witchcraft, then restored them. Then it's a little too white for a device THAT evil.
Well, we're used to see people in tv shows magnifying a clear picture of a killer from a cheap parking lot survilance camera. So, in a certain way, this doesn't surprise me....
I was going to mention that endless zoom into webcam pictures (I think it started in Blade Runner, which was at least a sci-fi, not a supposedly realistic police procedural).
Then I thought it was too geeky, even for me...
I used to like reading this blog, that is before I started reading the comments.

1) Converting audio to a greater bit depth/resolution will not give you better audio. Never. The absolute best result you can hope for is transparency on the conversion.

2) You can't hear the difference between 44.1/16 and 96/24. Even people with amazing hearing can only hear up to around 18Khz (I'm being very kind here) and ambient noise destroys any potential (read: ridiculous) dynamic range increase. Who here listens to music in an anechoic chamber and wishes to be taken from total silence to the threshold of pain? And why would anyone think that Nyquist was wrong?

3) You do not get better quality from transcoding an MP3 to a .wav. Just bigger files. I can't believe anyone here actually posted this garbage.

My honest guess is that this thing us a compressor and some kind of harmonic enhancer (bass boost) device. And if it is, $80 is about $79 too much.
I find this much easier to understand when I think in terms of digital images.
Imagine a high-res uncompressed digital image of a book page, 10,000 pixels square. (Imagine this is the equivalent of a 24/96hz digital recording)
Reduce that resolution to 1,000 pixels square. (Imagine this is the equivalent of downsampling it to 16/44)
Then compress that image by making it into a jpeg - you'll be able to see some visible artefacts in the file. (This is turning the .wav into an mp3)

Now try to do the process in reverse. What do you get?
The compression artefacts won't disappear just because you turn the .jpg into a .bmp.
And when you re-size the image, each pixel in the smaller image is just duplicated 100 times in the bigger image. You end up with a really big, smudgy, pixellated image, rather than a small, smudgy, pixellated image.

If you wanted to improve your big, smudgy, pixellated image, you could probably apply a bit of blur, to conceal the huge 10x10 pixel blocks and some of the artefacts. Maybe if you boosted the contrast a bit, and applied a sharpen filter, it might look better from a distance.
But the image details are lost forever. If you couldn't read the compressed small picture, you won't be able to read the 'uncompressed' big picture.
...the chart is brilliant - experience on the 'Y' axis? Total Brass Eye man.

could be a nice effect to put in the chain of distortion pedals :)
Creative = turgid shit encrusted in sparkly glitter...
No thanks, I'll stick to my quantum purifiers:
"An experience beyond studio qaulity"? So, studio quality is substandard? They got it wrong? The claim is bogus to begin with but when they claim that they can not only improve CD 'quality' to beyond what you'd get in the studio but also MP3 you just know that someone has decided to just make shit up. It's lossless! You lose information! It aint coming back and you absolutely can't go beyond it. the graph shows CD and MP3 at the same level after this process! Wait, I've got! They've put an opening into another dimension in the box and it's feeding directly from the studio when the recording was made! But no, I haven't got studio quality monitors at home set up for optimum listening. So, even if i did it wouldn't be as good. I'm starting to think that this is a joke and they'll pop up soon and say it was actually a set-up to get eveyone talking and then they'll launch the real product.

I don't think anyone has mentioned it yet but what about the HUGE difference in "experience" quality between CDs and MP3s. I'm assuming they meant 32kbs or something.

I also think that when they say studio they are referring the the TV show "The Actor's Studio". Because that show is really annoying. Listening to your iPlod with this sonic-monoknob-purification-transconductifier would be a much better experience than watching that show.

I still like Creative but find that they oversell products in the x-fi series.

I heard this one guy in Korea tried one of the prototypes out and when he played Cliff Richard's "Wired For Sound" through it, he glimpsed God for less than a second, then his speakers went on fire.

But get this - the fire was green. Spooky, eh? Probably best not to dabble in this stuff.
There is a clear difference between 44.1/16 and 96/24
Especially when using plugins (pitch shift?)

Ohh and nyquist was describing a perfect world.. just fyi.

Other than that I totally agree with you.. Its just a limiter or something.
All the algorithm talk is nonsense
re: "Experience on the Y-axis" --

What are the units? :)
"studio quality" is not a measurable thing.
To put it on a bar graph like that is ludicrous.
You can't hear the difference between 44.1/16 and 96/24. Even people with amazing hearing can only hear up to around 18Khz

What are you talking about? 18kHz, yeah, fine, but what's that got to do with the sampling rate? Nothing. And then you start howling about dynamic range. Garbage again.

But at least you got it right about converting mp3 to wave. I laughed out loud at the idiot who thought he can improve the sound quality that way. Because when you listen to the mp3, what you're hearing is precisely the same 44.1 kHz 16-bit audio -- bit for bit, this is a purely deterministic deal here folks -- as you'd be saving to a .wav file. Saving precisely the same zeroes and ones to a file before you play them back will not change anything. Retard.
Actually if you *in theory* had some offline codec that did a better job of decoding it, that for some theoretical reason couldn't run in realtime, then you might have better results. In reality though the math required to do FFT transformations are so well implemented in modern codecs, compilers & cpu's that we get high quality mp3 conversion in realtime for fractions of a percentage in your 'cpu meter'. That's pedantic though innit, as you definately won't get any quality increase out of converting to pcm audio with any decent mp3 codec.

"When it converts to 24-bit/96kHz, it has some "intelligent algorithm" for interpolating the values between 44.1 kHz samples. So you are getting smooth waveforms when you upsample, instead of it being squared off, and the interpolated values are shaped (so it isn't just a linear change or simple curve)."
This is actually a Slew Limiter, otherwise known as the Lowpass Filter. Every AD or DA stage has one. If you're 'smoothing' out artifacts that are occuring below the nyquist limit then why bother with the up conversion in the first place?

Imo I agree with Tony that they're most likely doing simple 'enhancement' of the audio plus perhaps eq curves. Cut mixer features off your existing hacked Emu chips and create a whole slew of consumer devices, including using the Emu dsp for simple enhancements. If they're doing dynamics compression I would guess it's most likely limiting, as modern music tends to be highly compressed already but limiting would help when a 'smile' curve is added. Unlikely they're doing esoteric phase manipulation like SPL, as you can't phase shift enough to restore what isn't there. Most likely a divided/companded style "big bottom" and gated noise style "enhancer" device like the Aphex devices as you'd be synthesizing extreme frequencies from existing material.

As for the remarks of Jeremy Cox, doing your dsp functions in 24/96k does buy an increase in resolution, something that filters (eq) especially benefits from. Since the Creative/Emu consumer chips have long been able to work at 24/96 internally and the dsp here is going to be dedicated in the standalone device, why the heck not? Heck, it lets them market it as 'better than studio!" Since, of course, esoteric dsp functions that work at 24/96 are unheard of in the pro studio realm.
This isn't a case of fools being seperated from their money, its just ordinary people who don't know any better who are being cynically pumped by Creative. Again.

FWIW I can hear the diference between 16bit and 24bit, but have yet to give a toss.
The SI unit of experience is the Ooh. Not to be confused with the imperial unit, the Aah. No one's used Aahs for years...
In the end it was this chart that made my buy a new Soundblaster. Better than studio quality, especially with my poorly mastered music. :P
My current soundcard has only 4 Oohs experience. This one has like nine!

Tony, If you knew as much about digital sampling as you want us to think, you'd know that a higher sample frequency can make up for the errors caused by bit depth quantization.

Try 1 bit at 1 MHz :P
It does have a nice big knob on it, though...
"Well, we're used to see people in tv shows magnifying a clear picture of a killer from a cheap parking lot survilance camera. So, in a certain way, this doesn't surprise me...."

I have been working on an algorithm to do this. The trick is to use to extract information from as many preceding and following frames as you can then deconvolve the whole shebang into a single frame.
The hard part is tracking the part of the image you are interested in, and you need a reasonable amount of frames to play with for it to work. Oh, and there needs to be no motion blur (as it works at the moment anyway).

If you had a recording made with a single mic of an instrument that only played a single identical repetitive sound then it might be possible to apply this technique to audio, and reconstruct that single note with more information than Shannon's theorem apparantly allows.

Somehow I doubt Creative are doing this. :)
I work on extrapolation algos (for terrain maps though). Feature recreation makes a good approximation of the data (involves a apriori knowledge of what the data was all about). We use fractals, non-linear dynamics to do this, though I very much doubt that this could be done on a $80 hardware, this not completely impossible.
Post a Comment

<< Home
Music Thing on Twitter
    follow MT on Twitter

    Music Thing Hits:
    Music Thing Heroes:
    Music Thing Friends:
    My music gear for sale
    DIY Modular Synth
    Matrix Synth
    Create Digital Music
    Analog Industries
    Boing Boing Gadgets
    London Video Production
    Wire to the Ear
    Palm Sounds
    Noise Addicts
    Retro Thing
    Analogue Haven
    Music Thing Massive
    About Music Thing:
    Send tips to Music Thing
    About this site
    Music Thing Massive
    RSS Feeds

    Problem with the ads?
    Please let me know