AlChuck
Well-known member
?????
M
MaxB
New member
AlChuck said:
I record with Cool Edit
D
darnold
New member
MaxB - What helps me support that you have no idea what your talking about is that fact that you arnt even making sense of what you are talking about! What are you talking about doing Asio in Cubase, Nuendo, etc? Alchuck said nothing of it.
But you can keep thinking that your SbLive can do 32/96 if you want. But your recording will never be like recording into a pro card or even pro-sumer card that can really record at 24/96. Infact you are probably just damaging the audio more than it is already damaged by running it through the SBLive card anyway.
Im done with this thread.
Danny
But you can keep thinking that your SbLive can do 32/96 if you want. But your recording will never be like recording into a pro card or even pro-sumer card that can really record at 24/96. Infact you are probably just damaging the audio more than it is already damaged by running it through the SBLive card anyway.
Im done with this thread.
Danny
Toki987
Rock Steady
darnold said:
I`m curious as to your level of experience in the mechanics of these soundcards. That last statement was reeking of uninformed bullshit. While techinically speaking of attributes the card does or does not represent, these attributes are relevant only in the circumstance with what you are trying to do and ultimately what result are you looking for. The efficiency of the actual bit rate in a/d conversion is quite sufficient for home recording use as a tool for production to simply do multitracking to see some ideas come to fruition, and providing of course that one is not planning on mastering the results from that card for commercial release. I would highly recommend it as an inexpensive, although there some better out there, entry level card.
D
darnold
New member
My point about damaging the sound was made because of the fact that the SB Live does have poor A/D converters in the first place. And trying to record through A/D converters with formats that the hardware doesnt support probably does more damage than good. Especially when the SB Live downsamples in the first place.
Now i do agree it is wrong of my to insist that he needs a pro-sumor or professional grade sound card, because yes all he may need is a SB Live. But that has not been the point of his posts, but the fact that he can magically do 32/96 sampling. I thought it would be good for him to know that the card probably isnt really doing 32/96 and could possibly be doing more damage to the sound then it is doing good. It seems in his sense of mind he is getting true 32/96 and would have no reason to upgrade. But in reality if he wanted to get a true 32/96 (if their are any cards that even support 32 bit) he would have to consider upgrading.
If the hardware could support such a format, why wouldnt Creative use that to their advantage? It would be a huge selling glitch.
The main thing is i apologize for coming across as arrogant and unrespective to your thoughts.
Danny
Now i do agree it is wrong of my to insist that he needs a pro-sumor or professional grade sound card, because yes all he may need is a SB Live. But that has not been the point of his posts, but the fact that he can magically do 32/96 sampling. I thought it would be good for him to know that the card probably isnt really doing 32/96 and could possibly be doing more damage to the sound then it is doing good. It seems in his sense of mind he is getting true 32/96 and would have no reason to upgrade. But in reality if he wanted to get a true 32/96 (if their are any cards that even support 32 bit) he would have to consider upgrading.
If the hardware could support such a format, why wouldnt Creative use that to their advantage? It would be a huge selling glitch.
The main thing is i apologize for coming across as arrogant and unrespective to your thoughts.
Danny
AlChuck
Well-known member
MaxB, look, the ADC in the card cannot give you better than 16-bit samples at 48 kHz.
I'm not doubting that you can change the sampling rate in software. All the software has to do is copy each sample, do that before where there was one sample at 48 kHz, you now have two -- the original and a copy -- at 96 kHz. And the 16-bit value is simply padded with zeroes.
The upshot is that the recording you think is being done at 32 bits and 96 kHz is merely the same data made to fit a different file format. It will sound virtually identical to the same material recorded at 16/48 because it is the same. You have not recorded at a greater resolution than the hardware can provide. The only result of saving your files in that form is a file many times larger (2x the number of samples and storing larger 32-bit values rather than 16-bit values for each sample) than it really needs to be.
What I said earlier about the larger bit depth used for processing, it's simply that when you do all the calculations involved with adding effects and such, you have less round-off error at the end of all the mathematical operations if you pad the 16-bit data with zeroes and perform the math with 24-bit or 32-bit values. But once again, this has nothing to do with the ADC and you are just plain stuck with 16 bits of resolution on the initial signal that you record.
I'm not doubting that you can change the sampling rate in software. All the software has to do is copy each sample, do that before where there was one sample at 48 kHz, you now have two -- the original and a copy -- at 96 kHz. And the 16-bit value is simply padded with zeroes.
The upshot is that the recording you think is being done at 32 bits and 96 kHz is merely the same data made to fit a different file format. It will sound virtually identical to the same material recorded at 16/48 because it is the same. You have not recorded at a greater resolution than the hardware can provide. The only result of saving your files in that form is a file many times larger (2x the number of samples and storing larger 32-bit values rather than 16-bit values for each sample) than it really needs to be.
What I said earlier about the larger bit depth used for processing, it's simply that when you do all the calculations involved with adding effects and such, you have less round-off error at the end of all the mathematical operations if you pad the 16-bit data with zeroes and perform the math with 24-bit or 32-bit values. But once again, this has nothing to do with the ADC and you are just plain stuck with 16 bits of resolution on the initial signal that you record.
Toki987
Rock Steady
It`s cool man, you didnt disrespect me. Just your opinion and it`s worthy. I was just standing in Live's corner to lend some support that it is definately a good choice for entry level when put into a cost/performace perspective. I use one myself, however I am aware of the limitations, especially the shortcomings in the a/d converters, although I dont feel that they damage the input material, maybe just not reflect the sonic accuracy in recording as would a higher end card reflect. I have seen the hype associated with the kx drivers and have tried them myself. The chore of installing the drivers and setting software up did not yield me any improvement that I could define. In fact the prior release actually increased the headaches. Quite possibly the new driver pack has some software trickery that may enable one to squeeze a bit more out of that chips performance as AlChuck stated, however I don`t believe any improvement can be made on the input side without a harware revision and its not likely that Creative would revisit that card and upset the economics of it in that marketing price range.
M
MaxB
New member
I can hear the difference between 48K and 96K recordings in my live (sound)... When I do editing with 96k recordings I'm very very happy with the final result. Why don't you all make some tests?
I have not time to learn technical stuff - I'm not interestes in that at all - the only thing I know it's that I'm happy working this way.
Currently I'm using a plugin under KX like a noise gate and the results It's no noise in line in at all... Not bad for a cheap game
card.
Can I post an example?
I have not time to learn technical stuff - I'm not interestes in that at all - the only thing I know it's that I'm happy working this way.
Currently I'm using a plugin under KX like a noise gate and the results It's no noise in line in at all... Not bad for a cheap game
card.
Can I post an example?
Toki987
Rock Steady
To me , if it sounds good its sounds good. 'doesn`t really matter if it was done on something homemade or cost megabucks, or anything in between. I think I`ll take a look at that new driver pack and see what it does. The technical part doesn`t matter, fidelity and dependability are the only issues I might have.
D
darnold
New member
That is totally fine if its working for you. But your claim of it being 32/96 is a little extreme.
danny
danny
M
MaxB
New member
darnold said:
I don't know if It's 32/96... What I know is that I can hear the difference between 48k and 96k recordings, just like I said before.
------------
Here's a sample:
AlChuck
Well-known member
You give us an MP3 to hear the difference with? And only one? What are we supposed to be comparing it to?
Give us WAV files, one at 32 bits, 92 kHz and one at 16 bits, 48 kHz...
Give us WAV files, one at 32 bits, 92 kHz and one at 16 bits, 48 kHz...
Last edited:
F
Fusion2
New member
yes, let's put this to rest once and for all, please...
i've been wondering about this issue for years now, even though, the results could be rigged/finalized easily...
thanks for the great clarifacation though, nice post...
i've been wondering about this issue for years now, even though, the results could be rigged/finalized easily...
thanks for the great clarifacation though, nice post...
J
Jim Y
New member
Cool Edit (Pro only? I've never worked with Cool Edit 2000) can record and save as 32bit from a 16bit input, but any advantage to this only comes with extensive editing, there is no more resolution in the original recording than a 16bit one.
It may also record at any sample rate you choose but this is derived by sample rate conversion. You can do this with MME, never with ASIO which must use the Audio sync clock of the soundcard - fixed at 48Khz for an SB-Live. This card is capable of carrying out sample-rate coversion itself but its analog converters are most definately fixed 16/48 in and out.
One advantage with the KX driver is that you can use the Rear outputs of the Live as the Front (swapped). The Rear outputs run from a seperate and higher quality Digital to Analog converter. Another reason why it might sound better with KX?
Recording at a higher bit depth or sample rate than the card is actually delivering definately won't increase any detail in the original recording, but it can help preserve that detail when extensive editing/fx is carried out on it. This is only at the expense of having much larger files and having to downsample to 16bit/44.1Khz when it's time to burn to CD.
The ability to "up-sample" your recording is done by Cool Edit and the MME interface, nothing special with the KX or any other driver or card. With MME, you can go the other way and record at 8bit/22Khz if you want!
It may also record at any sample rate you choose but this is derived by sample rate conversion. You can do this with MME, never with ASIO which must use the Audio sync clock of the soundcard - fixed at 48Khz for an SB-Live. This card is capable of carrying out sample-rate coversion itself but its analog converters are most definately fixed 16/48 in and out.
One advantage with the KX driver is that you can use the Rear outputs of the Live as the Front (swapped). The Rear outputs run from a seperate and higher quality Digital to Analog converter. Another reason why it might sound better with KX?
Recording at a higher bit depth or sample rate than the card is actually delivering definately won't increase any detail in the original recording, but it can help preserve that detail when extensive editing/fx is carried out on it. This is only at the expense of having much larger files and having to downsample to 16bit/44.1Khz when it's time to burn to CD.
The ability to "up-sample" your recording is done by Cool Edit and the MME interface, nothing special with the KX or any other driver or card. With MME, you can go the other way and record at 8bit/22Khz if you want!
AlChuck
Well-known member
That's what I'm talking about. A bit of a misnomer, though, to say it can record at a resolution of 32 bits. The recording is at 16-bit resolution; these values are converted into 32-bit numbers by the software for calculational purposes.
Again, loose use of the word "record." The sample rate of the recorded data is 48 kHz with this card. Sample rate conversion is performed by the software. I guess since it can be done on the fly maybe "record" is a valid word, it's a bit (no pun intended) gray, I'll admit.
M
MaxB
New member
I can't upload 96k wav files for comparison, too big in size, you know. (the mp3 was an example of the audio quality - Noise Level).
Here is two screenshots of the same recording. The pics shows the Spectral View in Cool Edit.
The First it's a 48K 16Bit recording.
The Second it's a 96k 32Bit recording.
The black sign shows a point where the difference it's evident.
Can you see the difference? The second picture shows a more dense spectre.
I know SBLive it's 48K but I've read something on kx forum... Well I don't remember... It's just like a doubled recording... The source is recorded twice at 48k to match the 96k sample rate.
It will be a good thing to ask Eugene Gavrilov (kxdrivers author).
http://web.infinito.it/utenti/r/riptide/48k16bit.jpg
http://web.infinito.it/utenti/r/riptide/96k32bit.jpg
Here is two screenshots of the same recording. The pics shows the Spectral View in Cool Edit.
The First it's a 48K 16Bit recording.
The Second it's a 96k 32Bit recording.
The black sign shows a point where the difference it's evident.
Can you see the difference? The second picture shows a more dense spectre.
I know SBLive it's 48K but I've read something on kx forum... Well I don't remember... It's just like a doubled recording... The source is recorded twice at 48k to match the 96k sample rate.
It will be a good thing to ask Eugene Gavrilov (kxdrivers author).
http://web.infinito.it/utenti/r/riptide/48k16bit.jpg
http://web.infinito.it/utenti/r/riptide/96k32bit.jpg
AlChuck
Well-known member
Sure you can. They don't have to be three-minute songs. A five-second snippet of the same source, recorded once at one setting and once at the "magic" setting, would suffice.
That's "spectrum." "Spectre" is a ghost; "spectra" is the plural of "spectrum."
About the pictures... first off, to me it looks like you labeled the files backwards -- the one labeled 48k16bit looks like it has finer resolution than the one labeled 96k32bit.
Secondly, even anything meaningful were really visible here, it tells nothing about how this data was derived. Namely that the data was 16-bit, 48 kHz, and the sampling rate was doubled, and the samples were padded with extra zeroes to make them 32-bit numbers. You have not recorded at a greater level of detail.
I didn't even mention this before but of course, when you play back your "32-bit, 96 kHz" file, it has to be downsampled to 16-bit, 48 kHz to come out of the DAC as audio again. So even if ther was some magic here, it would be washed away by playback through the card. Oh, I forgot, you think the software can make the DAC achieve specs beyond its capablity, so that argument won't phase you either.
All you are getting is bigger files. There is no difference that can be heard. If you think there is, it's your imagination.
Think of it like this: say you have a photo taken an 640 X 480 resolution on a camera.
In a graphics editing program, you can convert the file to twice the size. But the software cannot "know" what the right values are for the pixels it now has in the file but didn't record from the camera's imaging hardware. It can only make some sort of guess about what those pixel values might have been based on the original pixel.
The result is a bigger picture that takes more space to store but the original information is the original information. If there was some object that could have been seen in the picture had it been taken at twice the resolution, this object will not magically appear when you resize the photo in software.
M
MaxB
New member
Ok. These are two 5 seconds long waves of the same recording. Keep in mind that I'm not saying SBLive
can record at 96K, what I'm trying to say it's that I can hear (and see) the difference between one
recording and the "Magic"
one.Pardon for my bad english
Ok, now listen the two files below and take a look at their respective spectrum (between 15000 and 20000 Hz
you can see very well the difference).
No they are right. You are saying the 48k rec it's better than the 96k one?? But the 96k recording
shows more details in his spectrum.
Ehm I don't said this before: if you try to monitor through an Enhancer/Exciter with high settings
for the Hi End you can hear very very well the difference. 96k recording sounds to me more detailed
on the high range, more "airy". Do you own an enhancer? If so give it a try.
FLASHBACK: I Know SBLive It's a 48k 16bit card. I'm not saying it's a 96k 32bit card
http://web.infinito.it/utenti/r/riptide/48k16bit.wav
http://web.infinito.it/utenti/r/riptide/96k32bit.wav
J
Jim Y
New member
Perhaps there's no need for debate. If you peruse the KX website, you might find mention of a test program called RMAA or in full, RightMark Audio Analyser. This free program can test a soundcard by outputting a sequence of test tones and measuring what comes back. Although it isn't too scientific to test something with itself, it does give a useful comparative result.
Coming to those interesting pictures, I can't really see anything significant between the two at all! I am puzzled by the black hole in the 15-16Khz band in both. Have you done some heavy FFT filtering on this? It seems quite odd to have this deep cut in a signal which is otherwise full range in content.
Bands of cut and/or noise were quite evident with my old SBLive card which is one reason why I got rid of it. There really isn't anything driver design can do to repair this.
It might be handy to have the driver simulate 96k by interpolating extra samples from a 48k signal - if you need to record into an existing 96K multitrack project, but I don't see how the faked samples will do any better than the oversampling technique used by the cards hardware on a 48k signal.
I would also suggest you try the same comparison done with 32/96 and 16/48 with a 32/48 test. As far as I've read elsewhere, this is the best you can get out of an SBLive. 32bit because it gives more accurate results from edits and FX and 48Khz because it is the natural rate of the card and involves no sample-rate conversions - in or out.
Incidentally, check your recording software does not have dithering enabled on record. Then when you convert the sample type for CD or whatever, enable dithering for the best quality result.
I really would urge Max to try the RMAA program. He might find the frequency response of the Live card far from favorable. In contrast, even the very common Audiophile2496 has an excellent frequency response - as good as flat in fact. The difference in overall quality is obvious between these cards. Even soundfonts played back from VSampler thru an Audiophile sound better than thru a SBLive.
Having said all that, for the money, the SBLive 5.1 card is still better than on-board sound for someone on a budget and the KX project is something to applaud. Those behind it are most excellent people. I really would urge anyone making music with a SoundBlaster to try these drivers.
Coming to those interesting pictures, I can't really see anything significant between the two at all! I am puzzled by the black hole in the 15-16Khz band in both. Have you done some heavy FFT filtering on this? It seems quite odd to have this deep cut in a signal which is otherwise full range in content.
Bands of cut and/or noise were quite evident with my old SBLive card which is one reason why I got rid of it. There really isn't anything driver design can do to repair this.
It might be handy to have the driver simulate 96k by interpolating extra samples from a 48k signal - if you need to record into an existing 96K multitrack project, but I don't see how the faked samples will do any better than the oversampling technique used by the cards hardware on a 48k signal.
I would also suggest you try the same comparison done with 32/96 and 16/48 with a 32/48 test. As far as I've read elsewhere, this is the best you can get out of an SBLive. 32bit because it gives more accurate results from edits and FX and 48Khz because it is the natural rate of the card and involves no sample-rate conversions - in or out.
Incidentally, check your recording software does not have dithering enabled on record. Then when you convert the sample type for CD or whatever, enable dithering for the best quality result.
I really would urge Max to try the RMAA program. He might find the frequency response of the Live card far from favorable. In contrast, even the very common Audiophile2496 has an excellent frequency response - as good as flat in fact. The difference in overall quality is obvious between these cards. Even soundfonts played back from VSampler thru an Audiophile sound better than thru a SBLive.
Having said all that, for the money, the SBLive 5.1 card is still better than on-board sound for someone on a budget and the KX project is something to applaud. Those behind it are most excellent people. I really would urge anyone making music with a SoundBlaster to try these drivers.
M
MaxB
New member
Jim Y said:Perhaps there's no need for debate. If you peruse the KX website, you might find mention of a test program called RMAA or in full, RightMark Audio Analyser. This free program can test a soundcard by outputting a sequence of test tones and measuring what comes back. Although it isn't too scientific to test something with itself, it does give a useful comparative result.
Please check my previous message.
(RMAA) Well I think I'll give it a try...
Coming to those interesting pictures, I can't really see anything significant between the two at all!
Look at the circle... The 96k shows something (sonically) not present in the 48k rec.
I am puzzled by the black hole in the 15-16Khz band in both. Have you done some heavy FFT filtering on this? It seems quite odd to have this deep cut in a signal which is otherwise full range in content.
The Black Hole it's something like a "mirror effect"... The spectrum
it's mirrored and I don't know nothing about that. I was surprised like you the first time I saw this.
I would also suggest you try the same comparison done with 32/96 and 16/48 with a 32/48 test.
You are right. I think I will do this as soon as possible. Then I'll post the pics here. (Hey it tooks me 30 minutes to upload the waves above - prev. message
Incidentally, check your recording software does not have dithering enabled on record. Then when you convert the sample type for CD or whatever, enable dithering for the best quality result.
No my software don't have any dither on record.
About converting the sample type: Bob Katz say that the best thing to do is to leave the file as is... And if you convert from 96k to 48k you must disable dithering... The mastering house can provide the right dithering later.
Audiophile2496 has an excellent frequency response - as good as flat in fact. The difference in overall quality is obvious between these cards.
Ah Audiophile... I think this will be my next card. It's the first item in my letter to Santa Klaus. -3 months to Xmas Wow!!