Author Topic: Frequency range of PC soundcard (nVIDIA MCP61)  (Read 3446 times)

0 Members and 2 Guests are viewing this topic.

Offline jacksroadhouse

  • Stadium Superstar
  • ******
  • Posts: 1661
  • Good Vibes 48
    • Acoustic Passion Blog
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #30 on: April 05, 2012, 09:32:21 am »
Works in principle. It's just that my ears have fooled me big time once or twice in the past, so sometimes I'd like to do something else to verify what I'm hearing (or think I'm hearing).

If you're that bored, I've a lawn that needs mowing and some grouting...
Damn, I lost a bet with myself. At this time of year, I thought it would be the dirty dishes :)
Then again, redecorating someones garden sounds like fun. Who needs to know the difference between the lawn and and the flower beds, right?
You may want to think that one through first :)

Offline Majik

  • Stadium Superstar
  • ******
  • Posts: 2073
  • Good Vibes 124
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #31 on: April 05, 2012, 09:48:45 am »
I also learned recently on this very forum, that freqs above the human hearing range can actually have a negative impact on sound quality (something about speaker distortion, if I remember correctly). Which makes me wonder abput a thing or two...

It's very possible. There's strong evidence that frequencies outside the range of human hearing can have a detrimental impact due to negative effects it has when it passes through your hifi electronics (amp and speaker). Note usually music is recorded at very high resolution (96kHz or even 192kHz sampling rate). This contains an awful lot of ultrasonic noise, but whilst this is within the digital domain, this noise doesn't matter. One of the primary reasons for recording at such high sample rates is because the application of digital processing can create unwanted digital artefacts (noise) which are typically present in the upper frequencies relatively to the sample rate (if you've ever seen the visible "Moiré effect" on processed photos or on poorly chosen newsreader's ties, it's a similar phenomenon, but in the audio domain). By increasing the sample rate, you push this noise into the inaudible part of the spectrum (much like increasing the resolution of the camera or TV reduces the visible impacts of digital processing or bad ties).

The problem comes when you convert this to audio. Normally audio is rendered down to 44.1 or 48kHz sample rate by using sample rate converters. If you down sample to a higher rate (or don't downsample at all) then when you play these tracks, assuming your DAC doesn't apply any additional filtering beyond standard anti-aliasing, you will end up with a lot of this ultrasonic noise passing through your amp and speakers. It's inaudible in itself but it can (and probably will) still interact with the audible signal in unnatural ways within the electronics.

That was all rather off-topic, but one point that is relevant to the discussion: any audio DAC (digital to analogue convertor) including the one in your sound card will have "anti-aliasing" filters which are applied at around 1/2 the sample rate (so at around 22kHz for 44.1kHz audio). Thus the maximum your sound card will achieve will be about 24kHz if the maximum sample rate is 48kHz (which most sound card support). In practice I would be very surprised if any consumer grade sound card couldn't get very close to this: any major variations from this are really down to the quality of the filters, and modern digital filters are really very good. With modern components, it's very cheap and easy to build a DAC which outputs frequencies over 20kHz with 44.1kHz sample rate material.

I'm not a betting man, but I would bet money on any commercially available sound card today can achieve above 20kHz and, therefore, will have a frequency response well above that of adult human hearing (given most adults over 25 can't hear above about 16kHz). Consider that even modern cellphone electronics and speakers are capable of these frequencies: some school kids use a high-frequency "mosquito" ring tones on their cellphones which the adult teachers cannot hear.

What can vary, and will have an audible impact on quality, will be things like the linearity across the audible frequency range, the noise floor and THD. Onboard PC sound cards, in particular, tend to pick up a lot of noise and crosstalk from the computer mainboard.

Cheers,

Keith
« Last Edit: April 05, 2012, 11:39:22 am by Majik »
Guitars: PRS Singlecut S2, Fender Tele Lite Ash, G&L Legacy Tribute, Freshman Apollo 2 OCBX, Gibson SG Special P90
Amps: Bugera G5 Head, Boss Katana 100
All sorts of other stuff.

Offline jacksroadhouse

  • Stadium Superstar
  • ******
  • Posts: 1661
  • Good Vibes 48
    • Acoustic Passion Blog
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #32 on: April 05, 2012, 10:14:59 am »
Keith: off topic or not, that's really interesting. Thanks! I didn't know about those ring tones, love it! And: I found a couple of specs on another nVIDIA video/sound card from the same general family and it claims that they sample at 192 kHz (32 Bit), whoch fits right into our recent discussion about hires audio formats (in this case dazzling the customer with big numbers, I suppose). In or out wasn't mentioned though (maybe both). I found some strange phenomena with the headphones I mentioned, and what you wrote goes a long way in explaining that.

I know about the problems with noise and linearity. Looking at what cables they use, where they go and where they end up, I'm sometimes wondering how they keep the noise floor as (relatively) low as it is.

But still, I like to use what I got when it comes to satisfying my curiosity, and spend my money on stuff that really matters (still haven't bought that additional compressor).

Offline Majik

  • Stadium Superstar
  • ******
  • Posts: 2073
  • Good Vibes 124
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #33 on: April 05, 2012, 12:11:12 pm »
Keith: off topic or not, that's really interesting. Thanks! I didn't know about those ring tones, love it!

I'm glad you found it interesting. Incidentally the "mosquito" noise, which is around 17kHz, was originally designed as a device which could be used to discourage school kids from "hanging around" whilst not affecting adults. It, rightly, got a lot of bad publicity, and now the idea has backfired somewhat with the adoption of it as a "secret" ringtone.

Quote
And: I found a couple of specs on another nVIDIA video/sound card from the same general family and it claims that they sample at 192 kHz (32 Bit), whoch fits right into our recent discussion about hires audio formats (in this case dazzling the customer with big numbers, I suppose). In or out wasn't mentioned though (maybe both).

It's usually both, although that resolution cannot fully be realised in reality. Normally even the best A/D and D/A converters cannot usefully resolve audio beyond the 18th to 20th bit (at those levels the audio starts to get masked by the thermal noise in the components of the equipment). Are you sure it's 32-bit? Quite frankly anything above 24-bit A/D for audio applications is utterly pointless.

However, regardless of audibility, there are good technical reasons why you should sample at the highest quality you can, as well as doing any digital processing at high quality. You should keep the digital audio at a high quality: 24-bits fixed point is good enough for unprocessed audio samples. If you want to do any processing (mixing, fx, etc.) then 24-bit is generally good enough, but most DAWs now use 32-bit floating point (24-bit samples are automatically converted). Note that many processing plugins don't support more than 24-bit internally.

The big difference between analogue and digital is that analogue recording media (e.g. tape) normally has a very limited dynamic range (the useful space you can record into) and maximising it was important by keeping the recording levels high. With digital, this isn't a problem. With 24-bit resolution, you have so much dynamic range that you can sample something at a very low level and boost it up to a very high level in the digital domain without introducing significant noise or distortion. In fact you *should* aim to leave some headroom because unlike, say, analogue tape if you hit the recording peak even occasionally, it will have a dreadful impact on your recording: unlike most analogue media, with digital sampling the "peak" is an absolute brick-wall level that you should avoid at all costs.

When turning your audio into an "end product", convert to 44.1kHz or 48kHz at 16-bit making sure you use dithering. There's really no point in distributing it in a higher rate; the disadvantages outweigh the dubious benefits. Keep the hires masters though, for future remixing. Also make sure you have ample headroom in your final mix. Removing frequencies (even if they are inaudible) can cause very "hot" signals to start clipping. This is especially the case if the track is (or might be) converted to a compressed format like MP3.

Cheers,

Keith
Guitars: PRS Singlecut S2, Fender Tele Lite Ash, G&L Legacy Tribute, Freshman Apollo 2 OCBX, Gibson SG Special P90
Amps: Bugera G5 Head, Boss Katana 100
All sorts of other stuff.

Offline jacksroadhouse

  • Stadium Superstar
  • ******
  • Posts: 1661
  • Good Vibes 48
    • Acoustic Passion Blog
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #34 on: April 05, 2012, 02:01:46 pm »
Keith: so that's where that ringtone comes from?? They're apparantly still doing that sort of thing around here. A friend of mine is about to file criminal charges against a bank opposite the train station in his town (apparently it's a crime under German law). Hope he goes through with that, it should be a hoot!

And yes, it says 192kHz/32 Bit. This is one of them:
http://www.nvidia.com/object/mobo_gpu_features_benefits.html

The digital recording strategy you describe is pretty much what I'm doing (just for fun atm, I really need to do something real to go further in that department). It's interesting what you say about digital fx and such. I was wondering about that. Well, I usually record with my mixer's USB interface (I know, welcome to amateurville), and that does 48 kHz/16 Bit and that's it. But for my humle needs it's okay anyway.

I had that "hot recording" discussion with a friend of mine a little while ago. It was really funny, since his entire thinking is digital (in more than one sense), but he still preached that "as hot as possible" rule like it was gospel. When I told him to go buy himself a cassette recorder he got a little annoyed with me :)

I just tested my soundcard's and my audio driver's sample rate and bit depth and came up with at least 128 kHz / 24 Bit. That's the most that this tool can do, but it's pretty much above and beyond anyway.

So am I right in assuming, this means that the upper end of the freq range is at least a very, very generous 64 kHz (barring any kind of nonsense on the hardware level)?

Cheers

Jack

Offline Majik

  • Stadium Superstar
  • ******
  • Posts: 2073
  • Good Vibes 124
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #35 on: April 05, 2012, 03:10:46 pm »
And yes, it says 192kHz/32 Bit. This is one of them:
http://www.nvidia.com/object/mobo_gpu_features_benefits.html

Interesting! Pointless, but interesting!

Pointless because the minimum theoretical noise floor for an audio amp is -129dBfs (approx. 21 bit resolution). Very little real-world equipment gets near this, with -120dBfs (approx 20-bit resolution) being about the best. Most hifi equipment has a noise floor higher than 100 dBfs (approx 16-17 bit resolution).

32-bit resolution gives you a whopping -192dBfs noise floor.

Human hearing has an approximate dynamic range of 120dB, and the best acoustic rooms in the world may give you 130dB. Sampling audio at more than about 22 bit is pretty pointless as any bits beyond this don't contain information that is audible. But computers are naturally good at things in bunches of 8 so 24 bit is the appropriate size to use.

With 32 bit sampling, you would be capturing approximately 10 bits (roughly 60dB range) of inaudible noise. Harmless, but pointless.

NOTE: I'm using the 6dB per bit rule of thumb here. In reality this is inaccurate as you actually get more than 6dB per bit.

Quote
The digital recording strategy you describe is pretty much what I'm doing (just for fun atm, I really need to do something real to go further in that department). It's interesting what you say about digital fx and such. I was wondering about that. Well, I usually record with my mixer's USB interface (I know, welcome to amateurville), and that does 48 kHz/16 Bit and that's it. But for my humle needs it's okay anyway.

That's pretty standard. USB sound card are fine as long as people realise their limitations (especially for recording) but these limitations are often beyond what most people need anyway. The biggest problem with USB is the relatively poor performance of audio of USB. The impact of this is higher latency (which is really only an issue if you want to do software monitoring) and flakiness when recording multiple channels in parallel (which, if you are recording one or two channels at a time, isn't an issue). Also many USB audio drivers are limited to 44.1/48kHz and 16 bits: not ideal, but pretty damn good.

Quote
So am I right in assuming, this means that the upper end of the freq range is at least a very, very generous 64 kHz (barring any kind of nonsense on the hardware level)?

It should be, although I would be questioning the ability of the output drivers to support that. Handling signals well at 64kHz is going to be much harder than handling signals well at 20kHz. It wouldn't surprise me if it did support up to 64kHz, but with some significant distortion or attenuation compared to 20kHz. How much? unless they publish a data sheet, you'll probably need to do some testing (even most audio magazines rarely test much above 20kHz). You are also likely to be picking up lots of stray noise in those frequency ranges from the computer itself.

Obviously for audio apps this is irrelevant, but if you want to use it for scientific uses then it could be important.

Of course, given this is an audio card, it wouldn't surprise me if the onboard DAC was filtered to around 20kHz.I have heard an apocryphal story about a "hires" audio card which someone claimed to be better due to the higher frequencies, but when it was tested the on-board analogue output stages were filtering everything above about 22kHz regardless of input resolution.
« Last Edit: April 06, 2012, 10:26:27 am by Majik »
Guitars: PRS Singlecut S2, Fender Tele Lite Ash, G&L Legacy Tribute, Freshman Apollo 2 OCBX, Gibson SG Special P90
Amps: Bugera G5 Head, Boss Katana 100
All sorts of other stuff.

Offline jacksroadhouse

  • Stadium Superstar
  • ******
  • Posts: 1661
  • Good Vibes 48
    • Acoustic Passion Blog
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #36 on: April 05, 2012, 03:47:11 pm »
Well, at least it seems save to assume that 20 kHz and maybe a little highter is well within range for my soundcard. I can take it from there.

Offline TB-AV

  • Honorable Ex-Mod
  • All Time Legend
  • *****
  • Posts: 14966
  • Good Vibes 329
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #37 on: April 05, 2012, 03:52:39 pm »
You might want to read all of Dans white papers.

http://www.lavryengineering.com/index_html.html

Gone

Offline Majik

  • Stadium Superstar
  • ******
  • Posts: 2073
  • Good Vibes 124
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #38 on: April 05, 2012, 04:31:51 pm »
Guitars: PRS Singlecut S2, Fender Tele Lite Ash, G&L Legacy Tribute, Freshman Apollo 2 OCBX, Gibson SG Special P90
Amps: Bugera G5 Head, Boss Katana 100
All sorts of other stuff.

Offline TB-AV

  • Honorable Ex-Mod
  • All Time Legend
  • *****
  • Posts: 14966
  • Good Vibes 329
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #39 on: April 06, 2012, 12:01:41 am »

Quote
but he still preached that "as hot as possible" rule like it was gospel

He's about 8 to 9 years behind the times.

Tell your friend to listen to this.
http://www.youtube.com/watch?v=brdrKi6YugE&feature=channel


Then watch this...
http://www.youtube.com/watch?v=0nik1sJzQME&feature=related

In fact watch all of this guy's recording vids and keep an eye on --any-- of his meters. They are all the -15 to -12 range.


and this
https://www.youtube.com/watch?v=Ph1M3QZGku8&feature=related
Gone

Offline jacksroadhouse

  • Stadium Superstar
  • ******
  • Posts: 1661
  • Good Vibes 48
    • Acoustic Passion Blog
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #40 on: April 06, 2012, 07:38:26 am »
Thanks guys, that's good stuff (at least for me, my mate is more of a theorist). I didn't get through all of this, but it already filled in some blanks.

Offline TB-AV

  • Honorable Ex-Mod
  • All Time Legend
  • *****
  • Posts: 14966
  • Good Vibes 329
Re: Frequency range of PC soundcard (nVIDIA MCP61)
« Reply #41 on: April 06, 2012, 01:12:29 pm »
If you friend wants theory, tell him to look up the name Paul Frindle and see if he search out some of his teachings back around 2004-5. I think some of it has been removed due to various forum changes. But he was teaching the hows and whys of this stuff back then to the top engineers. Long before it was even re-written in all the home recording magazines and internet forums.

He kinda knows a thing or two about sound. http://www.proaudiodsp.com/about/
Gone

 

Get The Forum As A Mobile App