Hi all,
I'm a bit of an audio n00b, but still, I'm intending to play back ultrasonic frequencies (upto 32kHz) using a windows PC for my PhD project. To limit distortion, I'm looking into a system capable of playback at a rate of 192kHz. There's not that many cards capable of this it seems, and I'm currently eying the M-Audio Audiophile 192.
HOWEVER, it seems Windows 7 has the standard option of letting you choose the soundcard's sampling rate, between 44.1kHz and 192kHz. I don't really know what to make of this; this option seems to imply that the builtin audio on my motherboard is capable of operating at 192kHz sampling rates, even though there's only very few dedicated soundcards capable of operating at that rate... Is windows simply wrong here and presenting a standard option which my hardware doesn't support? And what happens when I set the sampling rate to 192kHz? Does it default back to the card's maximum rate (which I assume to be around 48kHz or 96kHz max..)?
Thanks for any help!
You could set the recording device to stereo mix, and directly record the output of your sound chip if operating at 192kHz. There are onboard chips which can do 192kHz, so likely Windows is correct.
You do understand that even though a card can play back 192 kHz files, that does not imply that it will output 32 kHz audio? The reconstruction filter may very well remove all ultrasonic frequencies.
You could set the recording device to stereo mix, and directly record the output of your sound chip if operating at 192kHz. There are onboard chips which can do 192kHz, so likely Windows is correct.
Thanks!
It's the ubiquitous Realtek HD Audio, which seems to be on every motherboard I've used these last few years.. I took the 'HD' as a bit of buzzwording to make it sound more interesting, but judging by wikipedia pages on Intel HD Audio it actually SHOULD operate at 192kHz..
Alright, I'm going to run some tests recording from one PC to the other; I can record using Audacity and check the spectrums using Spectrum Lab; or is this a very primitive idea?
You do understand that even though a card can play back 192 kHz files, that does not imply that it will output 32 kHz audio? The reconstruction filter may very well remove all ultrasonic frequencies.
No actually I did not; I should be able to find out running some tests.. Is there a way to change/bypass this reconstruction filter?
Is there a way to change/bypass this reconstruction filter?
I doubt it. There are web sites and tools that test sound cards, but I don't know if there are any tests of cards for their ultrasonic capabilities.
You do understand that even though a card can play back 192 kHz files, that does not imply that it will output 32 kHz audio? The reconstruction filter may very well remove all ultrasonic frequencies.
No actually I did not; I should be able to find out running some tests.. Is there a way to change/bypass this reconstruction filter?
No you cannot. Its not a discrete filter, its integrated into the design of the DAC. However, I doubt 32k will be blocked by the filter in most devices. 25k absolutely will not. Try and see.
You'll just have to experiment...
Usually if the hardware doesn't not support the sample rate, the drivers will do the conversion (which means you can get a 192kHz or 96 kHz file that's actually sampled at 48kHz). You can also usually get a 24-bit file recorded with a 16-bit soundcard.
The details are usually hidden from the user, but all of the applications, program material, and hardware/drivers work together (As long as they all follow the Windows standards) and that's an intentional feature of Windows. This is done for compatibiity. You'd like the ability to play a 24-bit file on a computer with a 16-bit soundcard. .. You wouldn't want the display of a picture to fail just because the image resolution is higher than your monitor resolution, or fail to print because your printer resolution is too low. (Compatibility is probably more important for playback than recording, but it works both ways.)
Intel HDA controller built into the chipset supports 192kHz. Recent motherboards have Intel HDA codecs which almost often support 192kHz. As a result integrated soundcards of majority of recent motherboards support 192kHz.
Now to the question of the DAC filter.
M-Audio Audiophile 192 uses AK4358. Its datasheet http://www.datasheetcatalog.org/datasheet2...7eldzgd8y7y.pdf (http://www.datasheetcatalog.org/datasheet2/a/0sgftyte7dclsklc97eldzgd8y7y.pdf) gives parameters of the digital filter on page 7 - flat response at least up to 80kHz for 192kHz fs. Page 33 shows recommended analog filter for the PCM mode - cut-off frequency at 90kHz. Such filter makes sense since the actual conversion samplerate is 384kHz, giving the 2nd order analog filter plenty of headroom up to the nyquist frequency of 192kHz.
The same analysis can be done for the integrated soundcard, we just need to know which specific HDA codec is used (motherboard documentation, googling, booting linux).
"To limit distortion" at 32kHz, there should be no problem with any decent soundcard that handles the standard 88.2kHz and 96kHz.
I have two Audiophile 2496s purchased in 2001. Testing some years ago revealed that program generated sweep tones up to the 96kHz limit (i.e. 48kHz) could be played on one PC and recorded on the other (D to A to D) quite well and true, with only a little loss of amplitude at the highest frequencies.
To limit distortion, I'm looking into a system capable of playback at a rate of 192kHz.
To elaborate on what AndyH said, using 192kHz rather than 96kHz will
NOT help with avoiding distortion at 32kHz, and may hurt.
Everything in Monty's explanation of why high sampling rates are counterproductive (http://xiph.org/~xiphmont/demo/neil-young.html) still applies to you- especially note the section on intermodulation distortion (http://xiph.org/~xiphmont/demo/neil-young.html#toc_1ch)- except with all references to 20kHz (the limit of human hearing) replaced with 32kHz (the limit of what you're interested in).
Use 96kHz (or 88.1kHz) instead.