[germanium]

So far all the free or oem supplied software I tried has 24 bit audio containers but only 16 bit internal code. This is very audible if you record at very very low levels then boost back up to low but audible levels. This low level distortion can make the resulting sound mildly grainy in worst case scenerio.

In true 24 bit recording the sound should be able to go super soft without audible distortion. Most analog sections have enough noise to dither the input of a true 24 bit ADC so the resulting file should have clean audio well into the noise floor just like properly dithered 16 bit but with a much lower noise floor. This is why the early CD transfers done from analog tape sounded so much better than the early digital ones. The tape noise served to dither the converters both at the recording end & at the playback end.

Even though Sonar LE outputs a 24 bit file it contains a dithered 16 bit file not the full 24 bit file. It is clear to me from my testing & I also found the setting for this in Sonar LE. It would not let me change that setting to 24bit as there was no 24 bit option there & it would not let me turn off the dither.

This setting is different from the one in the audio options. It is found under tools/change audio format. In litening to my test recording it is very obvious that it is only outputing the files at this settings resolution. So I still long for the true 24 bit recordings that my Xonar essence STX card is capable of.

 

[leeperry]

wavelab?

anyway, true 24 bit is not really possible on consumer stuff, you get 18 at best

and these CMI8788 cards only have one 24Mhz clock, I think they resample everything in 96KHz

 

[germanium]

Quote:

Originally Posted by leeperry /img/forum/go_quote.gif wavelab? anyway, true 24 bit is not really possible on consumer stuff, you get 18 at best and these CMI8788 cards only have one 24Mhz clock, I think they resample everything in 96KHz

The essence STX card is truely capable of good response up to the 192KHz sample rate limit (-3.5db @ 90KHz audio) so obviously it is not resampling 192KHz down to 96KHz sample rate.

Concerning the bit resolution of the ADC on this card is capable of real 20 bit reslution & comes very near to meeting that in actual tests. 118.5 db actual dynamic range with is only 1.5db off the target for 20bit resolution which is 120db. Thats taking into consideration all noise of the analog circuitry as well as the DAC. 18 bit would only yield 108db dynamic range.

24MHz is sufficient for 192KHz recordings since most DACs & ADCs drop thier oversample rate from 128x to 64x when operating at 192KHz which may be why the is a very slight reduction in performance at 192KHZ sample rate. This not really anything to worry about though. Only distortion is significantly affected & that is still way below audibility.

 

[leeperry]

oh sure, well it will resample everything in 48kHz multiples I believe, so 192kHz is entirely doable...I didn't explain myself properly. They simply have no 22Mhz clock available.

anyway, wavelab might be worth a shot...this is the real deal when it comes to audio editing/recording, if you're still not happy with how it fares out...you might be better off looking into the (poor) ASIO implementation of those DSP drivers than on whatever audio recording app not working properly.

 

[ROBSCIX]

Quote:

Originally Posted by germanium /img/forum/go_quote.gif So far all the free or oem supplied software I tried has 24 bit audio containers but only 16 bit internal code. This is very audible if you record at very very low levels then boost back up to low but audible levels. This low level distortion can make the resulting sound mildly grainy in worst case scenerio. In true 24 bit recording the sound should be able to go super soft without audible distortion. Most analog sections have enough noise to dither the input of a true 24 bit ADC so the resulting file should have clean audio well into the noise floor just like properly dithered 16 bit but with a much lower noise floor. This is why the early CD transfers done from analog tape sounded so much better than the early digital ones. The tape noise served to dither the converters both at the recording end & at the playback end. Even though Sonar LE outputs a 24 bit file it contains a dithered 16 bit file not the full 24 bit file. It is clear to me from my testing & I also found the setting for this in Sonar LE. It would not let me change that setting to 24bit as there was no 24 bit option there & it would not let me turn off the dither. This setting is different from the one in the audio options. It is found under tools/change audio format. In litening to my test recording it is very obvious that it is only outputing the files at this settings resolution. So I still long for the true 24 bit recordings that my Xonar essence STX card is capable of.

What about Adobe Audition? Your having the trouble with just ASIO or any drive set you use?
I think it is some type of config or driver issue not really the recording software itself. I am guessing though because I cannot test.
Have you considered sending an email to ASUS and asking them? I would like to help with testing but I am not using a STX...
I am interested in this issue though. Keep me posted. I will see if I can use my contacts and may be able to get some further informaiton for you.

 

[germanium]

Quote:

Originally Posted by ROBSCIX /img/forum/go_quote.gif What about Adobe Audition? Your having the trouble with just ASIO or any drive set you use? I think it is some type of config or driver issue not really the recording software itself. I am guessing though because I cannot test. Have you considered sending an email to ASUS and asking them? I would like to help with testing but I am not using a STX... I am interested in this issue though. Keep me posted. I will see if I can use my contacts and may be able to get some further informaiton for you.

Adobe Audition definately out of my price range. Tried Sony Sound Forge It suposedly lets you turn off the dither but doesn't really. On the 24 bit setting it sounds just like the 16 bit setting with dither supposedly turned off. If you turn on the dither the noise gets 3db louder in 16 bit than it is on 24 bit or 16 bit with dither turned off. This tells me that dither is always being used whether it is supposedly turned off or not. Even in the 24 bit setting. There should be a larger difference in the noise floor when turning off dither. It should drop back to the inherant noise of the ADC which is extremely low. about 25-30 db lower than the noise of the dithered output from the programs in question.

Wave Lab is 32 bit Windows compatable only. I run 64 bit Windows Seven.

 

[germanium]

Finally got both Audacity & Sound Forge to work properly as in 24 bit recording. Some how I had changed the settings in windows seven sound properties & I found them @16 bit 48KHz changed it back to 24bit 96KHZ & all is well. Sonar working well now as well

Full recording capabilities now as far as the ADC is capable but still need to replace some capacitors on the record circuit 4-47uf & 2-22uf electrolytics. The 2 I'm most concerned with are the 22uf caps as the seem to be a bleed for a negative feedback loop. This is the worst place for low grade electrolytics I've found. This is where they do the most damage to the sound & there is some slight degradation in the record circuit. You have to listen close but the massed voices are definately blurred to a small degree by them.

would like to try oscon or other solid state caps in this application. I have some 12uf 100volt metalized films but they are much too large physically to use there though otherwise they would be a perfect fit electronically speaking for this application.

 

[ROBSCIX]

Ah, the settings in the audio control panels. I though you were using XP or Vista...
I was thinking about those settings for Vista but figured the software would bypass any such settings.

When are you planning on modifying the recording channels?
I hope you posts some pictures and details.

 

[leeperry]

yes, it's really dumb that they force you to set a fixed sample rate in the W7 audio config

on one end they give you WASAPI, on the other they force you to SRC everything when it's not WASAPI...moronic coders, my signature never applied so well

the envy24 drivers on XP have an "auto" sample rate option...but it's gone on W7, as the system drivers don't seem to be allowed to change it by themselves anymore.

I don't recall seeing anything like this on Vista..XP when trimmed down is really a dream OS, fast/rock stable/bloat-free...they'll have to drag me by the hair to upgrade

 

[leeperry]

having second thoughts, a nice combo would be to set the Asus drivers GUI to 96kHz, W7 to 192kHz...and then play 44.1kHz files(mp3 if any possible) in KS/ASIO, so you'd get double SRC! craptacular

and if you're using Reclock in 48kHz, you get triple SRC

 

[germanium]

Made a mistake, those 2-22uf caps were to the AC97 codec which is used for the mike-in. The AC97 shunts this input to gound when not in use so it appears to act like a negative feedback bleed but it is not. Instead it goes direct to the AC97 codec. I decided to remove these caps & disable the mike on the Asus Xonar essence STX in the driver & use the onboard audio microphone instead which I think is better anyway. Of coarse that is the only thing that is better with the onboard sound. Signal is present on these caps even if you are using the line in which means a small portion is being bled to ground through the codec. not enough to significantly effect the overall output but enough to possably have a negative effect on the sound on the line-in.

When I substituted the 22uf caps with 12uf metalized films which have much lower loss & went straight to ground instead of through the codec to ground the sound started having soft ticks & pops indicating that by shunting the sound to ground may have a negative effect & stress the amps before them to a degree. I figured removing these caps altogether might be a good idea to improve sound. Imaging & soundstage seems to have improved but tonally it otherwise sounds the same when monitoring the line-in off my SACD player.

 

[germanium]

Quote:

Originally Posted by leeperry /img/forum/go_quote.gif having second thoughts, a nice combo would be to set the Asus drivers GUI to 96kHz, W7 to 192kHz...and then play 44.1kHz files(mp3 if any possible) in KS/ASIO, so you'd get double SRC! craptacular and if you're using Reclock in 48kHz, you get triple SRC

Resampling really isn't as bad as most people here are making it out to be. In fact there are distinct benefits to it in the very areas that people are assuming that resampling would be a negative.

Back some time ago some manufactures were using a chip to reclock the output from the SPDIF from on componant to anouther. Of coarse the manufactures didn't read the specs of the chip very well & assumed that that the jitter was being reduced when it wasn't. It turned out that the chip in question would only reduce jitter if you resampled the input to a different sample frequency. Otherwise it actually added some jitter as a result of the extra receiver chip.

Sample rate conversion nowdays can be done very very cleanly with a reduction of jitter. It is not unheard of to have distortion in the digital domain after resampling to be below -130db & sometimes below -140db. Both the X-Fi & Xonar cards are capable this low level of distortion when resampling. Consequently those looking for bit perfect sound may be actually getting worse sound as a result of thier insistance of bit bit perfect sound. I do understand that if you need DTS sound then you need that capability but most other situations you can possably get better sound by resampling.

 

[leeperry]

SRC adds harmonic distortion(easy to verify w/ WaveSpectra)..it's not a matter of SNR, it's a matter of unneeded additional THD/THD+N

better SQ due to resampling is not something I believe in

but I agree that the CMI8788 DSP only has a 24MHz clock, so indeed...resampling to a 48kHz multiple would help jitter, which would improve SQ...but it's mostly due to bad design. the Envy24 chips have two clocks to avoid this problem, and they forbid SRC:

 

[ROBSCIX]

Quote:

Originally Posted by germanium /img/forum/go_quote.gif Resampling really isn't as bad as most people here are making it out to be. In fact there are distinct benefits to it in the very areas that people are assuming that resampling would be a negative. Back some time ago some manufactures were using a chip to reclock the output from the SPDIF from on componant to anouther. Of coarse the manufactures didn't read the specs of the chip very well & assumed that that the jitter was being reduced when it wasn't. It turned out that the chip in question would only reduce jitter if you resampled the input to a different sample frequency. Otherwise it actually added some jitter as a result of the extra receiver chip. Sample rate conversion nowdays can be done very very cleanly with a reduction of jitter. It is not unheard of to have distortion in the digital domain after resampling to be below -130db & sometimes below -140db. Both the X-Fi & Xonar cards are capable this low level of distortion when resampling. Consequently those looking for bit perfect sound may be actually getting worse sound as a result of thier insistance of bit bit perfect sound. I do understand that if you need DTS sound then you need that capability but most other situations you can possably get better sound by resampling.

Not to mention the 8788 cards are bit perfect if set with proper sample rate. I agree that many make more of a issue out of it then it needs to be.

Always seem to be the people that don't really understand the topic.
IIRC, it is mentioned in the manual for the STX(X) the SRC engine uses high quality filters to drop any distortion below -140dB which is well below the audible range anyway.

 

[ROBSCIX]

Quote:

Originally Posted by leeperry /img/forum/go_quote.gif SRC adds harmonic distortion(easy to verify w/ WaveSpectra)..it's not a matter of SNR, it's a matter of unneeded additional THD/THD+N better SQ due to resampling is not something I believe in but I agree that the CMI8788 DSP only has a 24MHz clock, so indeed...resampling to a 48kHz multiple would help jitter, which would improve SQ...but it's mostly due to bad design. the Envy24 chips have two clocks to avoid this problem, and they forbid SRC:

Yes but as you have been told multiple times now, if you set the sample rate properly you get no resampling. As for the rest of your posting, I think you need to go and study the topic a bit further...