Yeah, I was hoping I could avoid having to get a dedicated soundcard - it's just that this seems like such a blatant software limitation that I find it hard to justify the expense when I know I won't be taking advantage of either EAX or the superior analog quality or anything else an Audigy or X-Fi could offer.
Can't get anything other than 48khz
- Thread starter 2008 IQ is unacceptable
- Start date
I actually use a podxt, a usb guitar amp/cab modeler which happens to work as an excellent headphone amp(strong headphone amp and excellent DACs) when gaming sometimes, if I have my headphones still plugged into it since with vista unless the game is EAX my audigy4 cant help offload work.
What all the other people said already, and here I thought my completely casual namedrop of Harry Nyquist would prevent that kind of crap
Dude, stop repeating yourself. I mentioned the minimum sampling rate required already, Nyquist is an old pal of mine
Again: that was not what I'm talking about, I'm talking about the frequencies contained in the original signal and not the sampling rate of the A/D converter.
Surely many people will hear the difference between a singnal containing only the filtered freq. below 22 kHz and say a downsampled 96 kHz signal. If you have a good rig and healthy ears anyway.
All that aside, even the cheapest, lowest mp3 will sound better if you can use 96 kHz instead of 48 at the output of your soundcard. Try it and listen for yourself, forget the theory.
No, it doesn't. A 64Kbit MP3 sounds absolutely no different at 44Khz, 48Khz, 96Khz or 192Khz. A 320Kbit VBR WMA might make a difference from 48 -> 96, but at the volume levels that I would need to play them at to make entirely sure, I can say that it doesn't matter enough to worry about it.
And the source I'm listening to is almost entirely orchestral encoded in D-D-D, so don't tell me that I'm somehow using shit for source...
I think you're entirely off your rocker, but not nearly as far off as IQ... Your reply sounds like another audio enthusiast bent on proving that somehow more expensive "upconverting" hardware somehow equates to better output -- exactly to your last point: forget the theory!
Yes, let's throw away the science that tells us exactly what's happening, and instead use our internal bias and need to feel better about ourselves by proving our superior worth as compared to those around us.
But we can't actually do that here _xxx_, because that's for the R&P forum...
For MP3, most encoders will lowpass filter the source file. The frequency of the filter depends on the bitrate you desire. So, it will toss out higher frequencies because of their questionable worth to make the audio simpler to compress. I'm not sure about other encoders but I have a feeling that others do it too.
The only way I can believe a 96 KHz -> 48 KHz resample could sound better is if the sample rate conversion process does something to the audio to make it subjectively more pleasing. It won't be as accurate as the original recording and you will lose the extra range offered by 96 KHz. There's no way to have > 24 KHz frequency with a 48 KHz sampling rate. But, subjectively, there are definitely things to consider for making audio more pleasing for people.
One thing I do wonder about is moving to 24-bit audio.
http://en.wikipedia.org/wiki/Quantization_(signal_processing)
http://en.wikipedia.org/wiki/Quantization_(sound_processing)
I used to have a better grasp on digital audio physics. It's all faded away. We need a digital signal processing pro to stop by or create a web site or something.
The only way I can believe a 96 KHz -> 48 KHz resample could sound better is if the sample rate conversion process does something to the audio to make it subjectively more pleasing. It won't be as accurate as the original recording and you will lose the extra range offered by 96 KHz. There's no way to have > 24 KHz frequency with a 48 KHz sampling rate. But, subjectively, there are definitely things to consider for making audio more pleasing for people.
One thing I do wonder about is moving to 24-bit audio.
http://en.wikipedia.org/wiki/Quantization_(signal_processing)
http://en.wikipedia.org/wiki/Quantization_(sound_processing)
I used to have a better grasp on digital audio physics. It's all faded away. We need a digital signal processing pro to stop by or create a web site or something.
Last edited by a moderator:
it's a sliding scale. Here's what LAME does, at least in an older version.
http://www.hydrogenaudio.org/forums/index.php?showtopic=13649
LAME will also resample the source for lower bitrates.
http://www.hydrogenaudio.org/forums/index.php?showtopic=13649
Code:
kbps lowpass
8, 2000
16, 3700
24, 3900
32, 5500
40, 7000
48, 7500
56, 10000
64, 11000
80, 13500
96, 15300
112, 16000
128, 17500
160, 18000
192, 19500
224, 20000
256, 20500
320, 21000the xfi crystaliser does something like that
heres the official marketing speil :
X-Fi Crystalizer instantaneously transforms your CDs and MP3s to crystal clear sound X-Fi Crystalizer™ analyses your audio file, identifies key elements that were lost or damaged during the compression process, and re-masters them using selective audio enhancement.
The mid-range and high-end (treble) becomes cleaner and clearer. Low-end (bass) frequencies are richer and more defined.
swaaye, have you seen a recent test of mp3 compression? i.e a spectrum analysis at various bitrates?
Please allow me to interject some of my own gathered knowledge and experience, since I'm a musician myself, and have dabbled with recording, and have picked up some pointers and things from some people in 'the industry'.
Firstly, the human hearing is limited to about 20-20.000 Hz (although the high frequencies drop off as you get older, most people over 20 probably can't hear over 17-18 KHz).
When a standard for digital audio was drawn up for new media like the CD, they settled on 44.1 KHz rather than the 40 KHz that would result from Nyquist.
The reason is that although you can represent a 20 KHz signal with a 40 KHz sampling rate, you get some aliasing near the top frequencies. Therefore they chose a sampling rate that is slightly higher, combined with a lowpass filter that removes all frequencies above 20 KHz (applied before A/D conversion, and again after D/A conversion to remove any aliasing). This results in a 'clean' 20-20KHz signal.
As A/D converters became faster and more accurate, they started sampling at 96 KHz, and then doing the lowpass 20 KHz filter in the digital domain. This was a cheaper solution and with better and more controllable output quality.
There was still a problem however... In the early days things sounded better when it was recorded and mixed on analog equipment, and then digitized. This had to do with the fact that digital mixing and processing is essentially a lossy process. As processors became more powerful, it became possible to do the recording, mixing and processing all at 24-bit 96 KHz (or 88.2 KHz, prefered by some because of less potential aliasing issues being a direct multiple of the CD frequency), which basically meant that you were 'supersampling'. The audio would only be filtered and converted to 16-bit 44.1 KHz at the end of the line.
Everyone pretty much agrees that 24/96 (or more) is the way to go in the studio (well, as far as people are convinced about digital in the first place, I've spoken with some producers that claimed that the analog equipment of the late 70s to mid 80s is still the benchmark, which I personally do not agree with), but not everyone agrees that you actually need more than 16/44.1 for the final medium.
At the least you could argue that if you don't downsample from 24/96 to 16/44.1, nothing can go wrong either, so it's just convenient... more of a what-you-hear-is-what-you-get approach.
I personally find 16/44.1 KHz quite acceptable, but I do require my processing at 96 KHz (you'd be surprised at how many semi-professional equipment is still stuck at 44.1 or 48 KHz). I also find DTS to be very good, if mastered and compressed properly (you can use multiband compressors and eqs to filter out certain 'problem areas' that could lead to compression artifacts). Even 128 kbit mp3 can sound pretty good when it is mastered properly (and a good encoder is used, some encoders sound massively better than others at the same bitrate. I use MP3Pro myself, which I find very good).
The most important thing to me is still the music itself. If the music is recorded properly, and there are no artifacts, then it's not that big a deal if the frequency range is not the full 20-20KHz, and the dynamic range might not be the full 16-bit or more (dynamic range is a joke anyway, in these days of loudness wars).
Heck, I play guitar, and I use a valve amplifier with speakers that probably don't go over 5-6KHz anyway. Doesn't mean it sounds bad... In fact, the limited frequency response is crucial to filtering out unpleasant frequencies in the signal. If you record directly from the preamp, the sound is really harsh. So you play through a speaker and use a microphone to record it. Or you use a circuit that simulates the speaker's effect on the sound. Digital modeling of amps/speakers is pretty good these days, and used on various recordings... but still there's nothing like the real thing.
Firstly, the human hearing is limited to about 20-20.000 Hz (although the high frequencies drop off as you get older, most people over 20 probably can't hear over 17-18 KHz).
When a standard for digital audio was drawn up for new media like the CD, they settled on 44.1 KHz rather than the 40 KHz that would result from Nyquist.
The reason is that although you can represent a 20 KHz signal with a 40 KHz sampling rate, you get some aliasing near the top frequencies. Therefore they chose a sampling rate that is slightly higher, combined with a lowpass filter that removes all frequencies above 20 KHz (applied before A/D conversion, and again after D/A conversion to remove any aliasing). This results in a 'clean' 20-20KHz signal.
As A/D converters became faster and more accurate, they started sampling at 96 KHz, and then doing the lowpass 20 KHz filter in the digital domain. This was a cheaper solution and with better and more controllable output quality.
There was still a problem however... In the early days things sounded better when it was recorded and mixed on analog equipment, and then digitized. This had to do with the fact that digital mixing and processing is essentially a lossy process. As processors became more powerful, it became possible to do the recording, mixing and processing all at 24-bit 96 KHz (or 88.2 KHz, prefered by some because of less potential aliasing issues being a direct multiple of the CD frequency), which basically meant that you were 'supersampling'. The audio would only be filtered and converted to 16-bit 44.1 KHz at the end of the line.
Everyone pretty much agrees that 24/96 (or more) is the way to go in the studio (well, as far as people are convinced about digital in the first place, I've spoken with some producers that claimed that the analog equipment of the late 70s to mid 80s is still the benchmark, which I personally do not agree with), but not everyone agrees that you actually need more than 16/44.1 for the final medium.
At the least you could argue that if you don't downsample from 24/96 to 16/44.1, nothing can go wrong either, so it's just convenient... more of a what-you-hear-is-what-you-get approach.
I personally find 16/44.1 KHz quite acceptable, but I do require my processing at 96 KHz (you'd be surprised at how many semi-professional equipment is still stuck at 44.1 or 48 KHz). I also find DTS to be very good, if mastered and compressed properly (you can use multiband compressors and eqs to filter out certain 'problem areas' that could lead to compression artifacts). Even 128 kbit mp3 can sound pretty good when it is mastered properly (and a good encoder is used, some encoders sound massively better than others at the same bitrate. I use MP3Pro myself, which I find very good).
The most important thing to me is still the music itself. If the music is recorded properly, and there are no artifacts, then it's not that big a deal if the frequency range is not the full 20-20KHz, and the dynamic range might not be the full 16-bit or more (dynamic range is a joke anyway, in these days of loudness wars).
Heck, I play guitar, and I use a valve amplifier with speakers that probably don't go over 5-6KHz anyway. Doesn't mean it sounds bad... In fact, the limited frequency response is crucial to filtering out unpleasant frequencies in the signal. If you record directly from the preamp, the sound is really harsh. So you play through a speaker and use a microphone to record it. Or you use a circuit that simulates the speaker's effect on the sound. Digital modeling of amps/speakers is pretty good these days, and used on various recordings... but still there's nothing like the real thing.
Last edited by a moderator:
Where have I mentioned upconverting or "more expensive"? You have a strong imagination. I'm talking strictly about the output.
I will bother replying to the stupid provocation: you are wrong. It IS about science, but only if you stop thinking as if your HW/SW was 100% ideal. There are many side-effects involved but that is too much for a forum post.
One more general thing, you need to learn how to communicate with less aggression. Maybe join an anti-aggression group or something.
Oh I dunno, maybe when you told us to up the playback rate to 96Khz? Or did you forget about typing that part?
It is still you who are wrong, not I. A 64kbit MP3 sounds no better being played back at 44khz than it does at 48khz or 96khz or 1Thz. It doesn't matter, 64kbps MP3's sound like crap regadless.
You're the one making the assenine claim, not I. A low-bitrate MP3 has information that is entirely missing, and no additional amount of upconverting or resampling is going to restore that data -- moving from the default 44khz playback to 96khz playback might as well be the same as moving to a 1Thz playback rate. It doesn't matter, the data is gone. You need something else other than a resample...
Oh, and the bit about aggressive communication style? LOL IRONY! So you must be this "pot" guy that all my kettle friends have been talking about...
Playback with 96 kHz is not the same as upconversion to 96 kHz. I'm talking about the former.
Again, look at electronics - interferences, reflexions in conductors, parasitic effects, signal quality issues and on and on. I'm talking purely about hardware issues here, not software or sampling issues.
Again, look at electronics - interferences, reflexions in conductors, parasitic effects, signal quality issues and on and on. I'm talking purely about hardware issues here, not software or sampling issues.
If you mean to say that there are many possible ways to upconvert signals to 96 KHz, and that there can be huge quality differences, then yes, you're absolutely right.
And if you're saying that playing a 44.1 or 48 KHz sample 'natively' sounds different from upsampling it to 96 KHz and then playing it, then yes, that pretty much follows from the above (although theoretically there exists a method of upsampling that sounds identical to a 'native' 44.1/48 KHz A/D converter).
However, if you claim that an upsampled 96 KHz signal ALWAYS sounds better than 44.1/48 KHz, then well... I think that's a purely subjective issue, and depending on too many factors (like which 'native' 44.1/48 KHz A/D converter you'd take, versus which 96 KHz upsampler etc).
I'd also like to add that many A/D converters have built-in resampling/interpolation/filtering circuits and the actual A/D conversion is usually done at a much higher rate than the actual input signal (known as oversampling), in order to minimize aliasing. So in practice you generally don't actually play a 44.1/48 KHz signal natively anyway.
From this it also follows that there are differences in sound quality and overall 'character' of the sound in different hardware (both in the algorithms used, and in the quality of the components, no A/D converter has a perfectly linear response, and there will always be some 'thermal noise' at the least, in the analog circuits). I think this is also something you meant to say, in which you are absolutely right.
And if you're saying that playing a 44.1 or 48 KHz sample 'natively' sounds different from upsampling it to 96 KHz and then playing it, then yes, that pretty much follows from the above (although theoretically there exists a method of upsampling that sounds identical to a 'native' 44.1/48 KHz A/D converter).
However, if you claim that an upsampled 96 KHz signal ALWAYS sounds better than 44.1/48 KHz, then well... I think that's a purely subjective issue, and depending on too many factors (like which 'native' 44.1/48 KHz A/D converter you'd take, versus which 96 KHz upsampler etc).
I'd also like to add that many A/D converters have built-in resampling/interpolation/filtering circuits and the actual A/D conversion is usually done at a much higher rate than the actual input signal (known as oversampling), in order to minimize aliasing. So in practice you generally don't actually play a 44.1/48 KHz signal natively anyway.
From this it also follows that there are differences in sound quality and overall 'character' of the sound in different hardware (both in the algorithms used, and in the quality of the components, no A/D converter has a perfectly linear response, and there will always be some 'thermal noise' at the least, in the analog circuits). I think this is also something you meant to say, in which you are absolutely right.
The built-in intrpolation/filtering circuits are usually a big part of that, yes (Audigy had issues there IIRC). But even lower level stuff like badly designed PCB or cheap connectors can mean a lot for the signal quality at different frequencies.
"Always" was meant as "with most usual devices out there", which are in majority some lower cost implementations.
"Always" was meant as "with most usual devices out there", which are in majority some lower cost implementations.