Audacity look s great and therefore I bought an expensive soundcard (Asus Essense ST )for analog output/recording.
But so far I am unable to record anything above 22 kHz, no matter if I use a sample rate of 192kHz.
When I play a 70kHz tone ( generated wit Audacity ) with VLC player it it heard as somthing like 7000 Hz or so.
I do have Linux too and if I get it to work as it should in either Windows XP or linux I will be happy.
Is there a hidden re-down-sample issue here or some anti alias filter that must be by passed?
Help is appreciated.
I have external test equipment (General Radio, Hewlet-Packard, Tektronix etc,) accurate out to 100KHz audio. What were you planning on doing?
The sound card analog is good to 90 KHz according to the web site. You can’t actually hear anything past about 15KHz or so (depending on how old you are). We had a production engineer where I used to work that could hear reliably out to 19KHz. That’s one in a row in decades of looking and he couldn’t go beyond that.
I suppose the cheapest way out is two cards and two computers and you could use one to test the other. I know of nothing in Audacity that would limit the high end response. Audacity routinely gets into trouble by having no natural limits.
like to do some measurements on speakers and amplifiers.
About the Asus website, don’t believe anything from Asus, as far as I know, nobody can get high resolution out there Essence cards
many are complaining about no support, no working drivers and Asus is telling plain lies to there costumers.
Normally it is simple to play and record at the same time in one computer, which I did, nothing comes in or out above 20kHz.
I Played 70kHz from another dedicated audio player and recorded the output, Audacity recorded nothing.
So I can be pretty sure nothing above 20 kHz comes in and a (very) little less sure about what is going out.
It seems that Windows and Linux for that matter, are re- sampling or have an anti alias filter which might be the problem.
I really hope someone ( like you ) can tell me how to fix that.
An high frequency played with VLC results in a much lower frequency.
Did you configure Audacity for 192kHz before generating that tone? Have you zoomed-in and checked the timing to make sure you really have a 70kHz tone? If you mathematicaly generate a 70kHz tone at the default 48kHz sample rate, you will get aliasing.
I wonder if it has something to do with the Windows drivers, but that doesn’t explain the aliasing… I’ve “heard” that Windows re-samples everything to 48kHz, maybe so that the digital mixer can mix everything at the same sample rate? But, that might only be true on older versions of Windows.
If the problem is the Windows drivers, you can try the ASIO drivers,. But, you’ll have to use the version of Audacity that supports ASIO, or another ASIO application.
\
BTW - if you are doing the usual thing and using your sound card for audio (for humans ), there are advantages in filtering-out any subsonic or supersonic signals (or noise).
The main advantage to a high-end soundcard is (potentially) lower noise, especially for recording. The other important specs are frequency response and distortion, but these are usualy better than human hearing even cheap soundcards. And of course, you can buy high-end audio interfaces with multiple inputs (for multi-tracking) and XLR mic inputs, etc.
if I use a sample rate of 192kHz.
There’s no harm in high-resolution recording. The professional recording standard seems to be 96kHz/24-bit, and many audio pros & audiophiles will tell you that “high resolution” is better. But, the guys who’ve done scientific blind testing have found that you can downsample a high-resolution recording to 44.1kHz/16-bit, and you can’t hear any difference.
I Played 70kHz from another dedicated audio player
Had you said “a dedicated test instrument” I’d be a lot happier. “Audio players” have to cut corners. 44100 is not good to 20 KHz like everybody keeps insisting. It’s only guaranteed good to 17 KHz (Nyquist 2.6) and it’s pretty much making it up for anything higher. That’s why sweeps at 44100 are never flat up at that high end. It’s just really good guesswork.
In order to not create audible problems, filtering is done to even out the dithering noise and errors. The filters will not pass further up than 20 KHz or maybe a little more. They can’t. They fall into the aliasing traps if they do. I’d be really surprised if simple audio players bother to switch the filters in and out as the sample rates go up. Why would they? Nobody is going to pay for services and electronics which will never be used and nobody can hear.
I think you’re stuck with actual test equipment to do these jobs. Shuffling consumer electronics in and out as proof of performance is an invitation to an impossibly complex Venn Diagram.
I friend of mine have a soundcard with ISA slot working under DOS, which works fine up to 50 kHz or so.
Asus says my card is supported by Windows XP and they provide a ( fake ) testreport which shows -3 dB at 90kHz.
Tried that doesn’t work either.
No problem with discussing that too but for now I like to stay to the problem, I cannot be the first which encountered this problem.
Had you said “a dedicated test instrument” I’d be a lot happier. “Audio players” have to cut corners. 44100 is not good to 20 KHz like everybody keeps insisting. It’s only guaranteed good to 17 KHz (Nyquist 2.6) and it’s pretty much making it up for anything higher. That’s why sweeps at 44100 are never flat up at that high end. It’s just really good guesswork.
My player works at 192kHz sample rate.
In order to not create audible problems, filtering is done to even out the dithering noise and errors. The filters will not pass further up than 20 KHz or maybe a little more. They can’t. They fall into the aliasing traps if they do. I’d be really surprised if simple audio players bother to switch the filters in and out as the sample rates go up. Why would they? Nobody is going to pay for services and electronics which will never be used and nobody can hear.
That is an different issue with lots of assumptions. No human can hear a sine wave of a fixed frequency of 50kHz, but it is jumping to conclusions that higher frequencies in music doesn’t make any difference.
I think you’re stuck with actual test equipment to do these jobs. Shuffling consumer electronics in and out as proof of performance is an invitation to an impossibly complex Venn Diagram.
Well I agree but as said, Asus claims my card can produce and record up to 90 kHz, and if it does I must be able to make that visible.
As far as I can tell now, it doesn’t! Now I want to know the reason why and solve that.
I’ll take a look at the Audacity-Linux site too now, pfff…
First, the usual caveat I don’t own an Asus sound card and have never actually seen one in the flesh.
No. XP will try its utmost to use the highest possible sample rate. IIRC the max was 100 kHz in the original release and that was raised to 200 kHz in Service Pack 1. Lots of detail here.
Here’s one guy that had a similar problem, but on Vista: “It appeared that 96 and 192kHz files were being downsampled by the STX to 48kHz, with spectral components higher than 48kHz aliased into the audioband; eg, a 40kHz tone was reproduced as 8kHz.”
His problem was resolved by a driver update.
Windows doesn’t do that, but many sound cards do. Especially Creative ones, many have a DSP that runs at a fixed 48 kHz and any sound not at 48 kHz is re-sampled.
On that note, I’ve read that the Asus Essense ST has a bunch of DSP effects available. I would think that to get true high sample rates, all DSP effects would have to be switched off.
From what I can glean from the interwebz the DSP-free mode is called “Hi-Fi” or something like that.
Since ASIO talks directly with the hardware, that rules out anything in the Windows sound system.
My guess is that this is a driver or setting issue .
Well, the reason I’m interested in high sample rates is that I’m not always doing the usual thing
I have an old XP computer that runs StereoTool to feed a transmitter with FM MPX audio. The MPX audio goes up to 75 kHz.
Sometimes I want to analyze the output from my transmitter. With the help of a hacked tuner, where I tap the output from the FM discriminator, I can record the raw MPX audio. The recorded audio is then used as input to Spectrum Lab.
I haven’t had to do anything special, I just set the sample rate to 176,4 or 192k in StereoTool and Audacity and go.
When I produce audio for human consumption, I’m never above 48 kHz.
I found this notice over at the JRiver Wiki
“The only way to prevent a Creative Labs X-Fi based card or Asus Xonar card from resampling all incoming audio is to use ASIO and the driver that came with the card. With these cards, WASAPI exclusive will not change the master clock of the card”
), there are advantages in filtering-out any subsonic or supersonic signals (or noise).