IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Which sample rate should I use w/ Xonar STX? Better performance at 96?, [was “Which default sample rate for my sound card ?” in General Audio]
extrabigmehdi
post Jun 29 2012, 14:12
Post #1





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



Hello,
before I was doing in my dsp chain in foobar:
sox set to 96 khz , followed by a big vst chain of dsp tweaked to my taste.
But I've realized that overall my dsp were performing worse than with 44.1 khz.

So now it's more like :
big dsp chain followed by sox set to 96 khz.

Why upsampling ?
My reasoning:

- the default sample rate of xonar stx after install is 96 khz

- I imagine that if I use different software, with different sample rate, they must be mixed to a single sample rate at one moment.
So I rather set a big sample rate on soundcard, and play music at that sample rate.

- the measurements at tech report website, indicate that the xonar stx perform better at 96 khz , than 44.1 khz.
for 44.1 khz : http://techreport.com/articles.x/17256/8
for 96 khz: http://techreport.com/articles.x/17256/9

- if I don't upsample , and set the default sample rate to 96 khz, I imagine that the upsampling
would be either handled by the windows mixer or the driver of sound card ?
So I rather use something that is well tested like sox.
I'm not sure if wasapi would bypass resampling by system.

- I can hear differences, but if you ask me to verify such claim, I don't know how it could be ABX ed.

Here are sample rate supported by soundcard:
44.1 khz, 48 khz, 96 khz, 192 khz

So which sample rate would you use , and why ? Since even tech report website indicate that the xonar stx perform better at higher sampling rate, is is pertinent to upsample using sox "on the fly" while playing music , or should I leave this work to the driver/soundcard/ windows mixer/ I don't know .
Does wasapi allow to bypass upsampling if the the sample rate set on soundcard doesn't match sample rate of played music ?

thanks
Go to the top of the page
+Quote Post
extrabigmehdi
post Jun 29 2012, 14:14
Post #2





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



oops sorry if I didn't post on hardware section. I didn't check, once more.
Go to the top of the page
+Quote Post
uart
post Jun 29 2012, 15:23
Post #3





Group: Members
Posts: 810
Joined: 23-November 04
Member No.: 18295



QUOTE (extrabigmehdi @ Jun 29 2012, 05:12) *
Here are sample rate supported by soundcard:
44.1 khz, 48 khz, 96 khz, 192 khz

So which sample rate would you use , and why ?

My Realtek based onboard audio performs best (according to my own RMAA loopback tests) at 48kHz, so yeah I also up-sample (44.1 to 48) in foobar. I'm not sure I can hear it, but there's a couple of dB measurable improvement so I figure why not.

You should try an RMAA loopback test at 48kHz and if it's as good as 96kHz (I'm guessing that it will be) then I'd say use that (48k).

This post has been edited by uart: Jun 29 2012, 15:23
Go to the top of the page
+Quote Post
uart
post Jun 29 2012, 15:35
Post #4





Group: Members
Posts: 810
Joined: 23-November 04
Member No.: 18295



QUOTE (extrabigmehdi @ Jun 29 2012, 05:12) *
- I can hear differences, but if you ask me to verify such claim, I don't know how it could be ABX ed.


To be honest, looking at those RMAA results I'd be surprised if you could. What type of differences are we talking about? From looking at the links I'm guessing the difference is something like 0.0012% THD versus 0.0010%. Nice soundcard btw, good linearity (low THD and IMD). smile.gif

This post has been edited by uart: Jun 29 2012, 15:36
Go to the top of the page
+Quote Post
lvqcl
post Jun 29 2012, 16:02
Post #5





Group: Developer
Posts: 3383
Joined: 2-December 07
Member No.: 49183



QUOTE
- the measurements at tech report website, indicate that the xonar stx perform better at 96 khz , than 44.1 khz.


Do they? Or they indicate that 16 bits (44.1/16) is better than 24 (96/24)?
Go to the top of the page
+Quote Post
Ceniza
post Jun 29 2012, 19:43
Post #6





Group: Members
Posts: 22
Joined: 30-May 12
Member No.: 100288



On my side I have a Xonar HDAV1.3 Deluxe connected to an A/V receiver via HDMI. I use foobar2000 for playback with the WASAPI plugin (which uses exclusive mode) on the digital output, and I only use very basic DSPs to do some channel manipulation (no EQ, no resampling, no anything that alters the signal). This way the A/V receiver gets the signal at its original sampling rate while skipping Windows' mixer.

In case you cannot use digital output, you should let your ears decide. If your ears cannot do that (maybe because they cannot tell the difference), and all scenarios perform smoothly, then follow your heart (or flip a coin). Still, give WASAPI a go.
Go to the top of the page
+Quote Post
extrabigmehdi
post Jun 30 2012, 05:32
Post #7





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



So, I finally did this famous RMAA Test for the first time (loopback test).
I wasn't sure which recording device to enable on system, I used "stereo mix":



So I tested all sampling rate using 16 & 24 bit depth.

It was confusing that the xonar stx allow to set a sampling rate with it's panel,
while windows allows to set also another sampling rate in the mixer, playback section huh.gif
Maybe there's a bug (both value should remain same, regardless of the interface used).

I found the results of rmaa a bit reliable, and although I tried to not touch the volume knob, I saw variations in the reported volume. Sometime I got too wrong result (maybe because I surfed a bit ?) so I've redone test.

Any here's a first serie of results:


And an other serie of results


I'm not sure there's a winning sampling rate, but I thought that 192 khz was more reliable.
Let me know what you think

This post has been edited by extrabigmehdi: Jun 30 2012, 05:35
Go to the top of the page
+Quote Post
uart
post Jun 30 2012, 09:28
Post #8





Group: Members
Posts: 810
Joined: 23-November 04
Member No.: 18295



QUOTE (extrabigmehdi @ Jun 29 2012, 20:32) *
I found the results of rmaa a bit (un?)reliable, and although I tried to not touch the volume knob, I saw variations in the reported volume. Sometime I got too wrong result (maybe because I surfed a bit ?) so I've redone test.


I notice that you're recording the "stereo mix", so another other sounds (eg synth or un-muted mic or line-in) would cause a problems. The one-off 78dB measurement certainly looks like an anomaly.

Apart from the anomaly, I'd say all the 48kHz multiples look very similar. Given a bunch of "equivalents" I'd go for the lowest common denominator, 48kHz.

Go to the top of the page
+Quote Post
skamp
post Jun 30 2012, 09:55
Post #9





Group: Developer
Posts: 1444
Joined: 4-May 04
From: France
Member No.: 13875



I'm not sure the "Stereo Mix" feature makes the audio go through the DAC and the ADC? I.e. did RMAA actually measure the sound card?

This post has been edited by skamp: Jun 30 2012, 09:56


--------------------
See my profile for measurements, tools and recommendations.
Go to the top of the page
+Quote Post
lvqcl
post Jun 30 2012, 10:49
Post #10





Group: Developer
Posts: 3383
Joined: 2-December 07
Member No.: 49183



QUOTE
while windows allows to set also another sampling rate in the mixer, playback section


...and (probably) also in recording section: Stereo Mix -> Properties -> Advanced.

Also: http://www.hydrogenaudio.org/forums/index....showtopic=86676
Go to the top of the page
+Quote Post
uart
post Jun 30 2012, 13:14
Post #11





Group: Members
Posts: 810
Joined: 23-November 04
Member No.: 18295



QUOTE (skamp @ Jun 30 2012, 00:55) *
I'm not sure the "Stereo Mix" feature makes the audio go through the DAC and the ADC? I.e. did RMAA actually measure the sound card?

I would have been concerned about the same thing - what exactly is stereo mix - however his results certainly look consistent with having been through the DAC-ADC chain. Traditionally the soundcard's "stereo mix" is an analog mix of the DAC output and any other available analog sources, including line-in, mic-in and possibly the synthesizer output. Personally I prefer to use an external cable (line-out to line-in).
Go to the top of the page
+Quote Post
extrabigmehdi
post Jun 30 2012, 13:22
Post #12





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



Yeah I meant that the RMAA test was a bit unreliable .

Each time I've set to same values the sample rate & bit depth :
1 - on xonar panel
2 - on playback device used , from windows mixer
3 - on recording device used, from window mixer
4 - the rmaa test , itself

@skamp
QUOTE
I'm not sure the "Stereo Mix" feature makes the audio go through the DAC and the ADC? I.e. did RMAA actually measure the sound card?


I don't know, maybe I've done something wrong ? Should I reconfigure
the recording device each time to match the sampling rate & bit depth ?

Here's how I 've done connection: between headphone output
and "line in". The xonar stx use the same connector for line in & mic.



@uart

QUOTE
I notice that you're recording the "stereo mix", so another other sounds (eg synth or un-muted mic or line-in) would cause a problems. The one-off 78dB measurement certainly looks like an anomaly.


I noticed before other anomalies, and I've redone few tests.

Also regarding the cable I used for loopback, I'm not sure if it is good. In both end of cable I've added a jack adaptor.
Go to the top of the page
+Quote Post
uart
post Jun 30 2012, 13:44
Post #13





Group: Members
Posts: 810
Joined: 23-November 04
Member No.: 18295



QUOTE (extrabigmehdi @ Jun 30 2012, 04:22) *
Here's how I 've done connection: between headphone output
and "line in". The xonar stx use the same connector for line in & mic.

Also regarding the cable I used for loopback, I'm not sure if it is good. In both end of cable I've added a jack adaptor.

Ok, I see you actually have made an external loopback connection. In this case you must select "line-in" instead of "stereo mix" as your recording source. If you use "stereo mix" then you don't actually need the loopback cable (the loopback is actually internal to the card).

QUOTE
I noticed before other anomalies, and I've redone few tests.

While using "stereo mix" can still work (at least on some cards), as you've found it can also record extraneous sounds which totally spoils the test. You need to get the external loopback cable sorted out and select line-in as the source.


Go to the top of the page
+Quote Post
extrabigmehdi
post Jun 30 2012, 16:19
Post #14





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



Ok , I've redone the rmaa test with "line in" , and I'm much more confident with results. The reason I haven't used "line in" option before, is because on rightmark website, they provide a tutorial for the xonar dx; and on that tutorial , they didn't use "Line in", but "Alt" (there's nothing with that name with the xonar stx).

On the windows mixer:



Results:



Let's redo the test, to check reliability:




If I interpret correctly results,
I think that for 24 bits, the soundcard perform best at 48 khz,
and for 16 bits, the soundcard perform best at 192 khz.
So a good compromise for general use, would be 96 khz.

Let me know what you think.
Go to the top of the page
+Quote Post
stephan_g
post Jul 2 2012, 17:03
Post #15





Group: Members
Posts: 197
Joined: 6-June 10
From: Bavaria
Member No.: 81240



You should be getting better 16-bit results still when setting the headphone out to "600 ohm", though this may be pushing the abilities of the input stage. An adapter cable for a line-out to line-in connection may yield better results.

Your 16-bit results appear to be almost entirely noise limited. Otherwise it will seem that distortion gets a hint worse at 192 kHz, and 44.1 kHz shows the reduced dynamic range usually observed with these cards (I am guessing that the master clock is generated with a somewhat noisy CMOS PLL rather than devised by crystal clock division in this mode).
Go to the top of the page
+Quote Post
extrabigmehdi
post Jul 2 2012, 18:22
Post #16





Group: Members
Posts: 408
Joined: 15-August 09
Member No.: 72330



QUOTE (stephan_g @ Jul 2 2012, 16:03) *
You should be getting better 16-bit results still when setting the headphone out to "600 ohm", though this may be pushing the abilities of the input stage.



I've been told that changing the headphone impedance setting in the panel of xonar stx , does nothing special : it just define a new volume limit.
I've "optimally " set the volume by following the rma test indications (an indicator becomes red if volume is too low, and also if too high).

This post has been edited by extrabigmehdi: Jul 2 2012, 18:23
Go to the top of the page
+Quote Post
dhromed
post Aug 31 2012, 09:02
Post #17





Group: Members
Posts: 1314
Joined: 16-February 08
From: NL
Member No.: 51347



QUOTE (extrabigmehdi @ Jun 30 2012, 06:32) *
I saw variations in the reported volume.


I don't know about the STX, but my DG has a limiter/compressor that activates automatically when your set the volume control above 50%, which might influence your measurements. The result is weird wobbly volume, as if a tiny gnome were randomly changing the knob a little.

This auto-compressor is different from the described SVN feature, which you can actually turn off.

QUOTE
I've been told that changing the headphone impedance setting in the panel of xonar stx , does nothing special : it just define a new volume limit.

This is correct.

This post has been edited by dhromed: Aug 31 2012, 09:03
Go to the top of the page
+Quote Post
YellowOnion
post Oct 22 2012, 23:07
Post #18





Group: Members
Posts: 29
Joined: 9-April 12
Member No.: 98589



I did my own loopback testing via sox after I was hearing some audible tones in a frequency sweep, turns out my soundcards (Xonar DS) built in resampler Aliases badly.


Heres a clearer image with a 0 to -20dB range



I assume the DSP code is the same for all models?
Go to the top of the page
+Quote Post
Kohlrabi
post Oct 23 2012, 00:40
Post #19





Group: Super Moderator
Posts: 1050
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



Yet another hint to avoid ALIASUS hardware like the plague.


--------------------
Ceterum censeo Masterdiskem esse delendam.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 22nd September 2014 - 16:42