 |
|  44,1khz vs. 48khz - Can you hear a difference? [MUSIC] | |
|
 |
Atlantis
  Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
 |
Xtatic wrote on 10 Dec. (23:10) :
Actually I knew that haha.. Reading for exam and shit. But quantizing error and such rarely happends, and with 44,1 you have the little extra you need in case of quantizing errors.. But people here speak of it as if they can hear a difference in quality, as in comparing it to 128kbps to 256kbps.. Maybe I misunderstood..
|
Quantisation error happens every time you perform a calculation on digital audio. If you're processing in 16 bit, the rounding errors will accumulate as a graininess in the sound. It also happens when you take a 16 bit sample, process it, and save it as a 16 file again. That's why processing is often done much higher (32 or 64 bit float), and you dither the output down to 16 bit at the end.
And the sample rate has nothing to do with quantisation error.
And I'd say the difference is like comparing 128 kbps to 160 kbps, or 192 kbps to 224 kbps. It's really quite easy to hear if you know what you're listening for, but at the same time it's easy to accept it as part of the sound too, and yet it's what makes digital sound digital and analogue sound analogue.
So render at a high sample rate to minimise aliasing distortion if you have a lot of high frequency content, and resample down to your final sample rate (using a lowpass filter to prevent these high frequencies from becoming low frequency distortion) at the end, but before dithering. As I said before, it depends on your equipment too, and how certain synths/resampling algorithms handle frequencies above the Nyquist frequency.
My mind is drifting... |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
|
Atlantis wrote on 11 Dec. (6:52) :
And the sample rate has nothing to do with quantisation error. |
Someone please correct me if I'm wrong because I don't think I'm getting it. Wikipedia's dither page says:
|
Another plausible solution would be to take 4.8 and round it so that four times out of five it rounded up to 5, and the other time it rounded to 4. This would average out to exactly 4.8 over the long term. Unfortunately, however, it still results in repeatable and determinable errors, and those errors still manifest themselves as distortion to the ear (though oversampling can reduce this). |
How can oversampling reduce this? And if so, that would mean quantisation is related to the sample rate, as outlined on the oversampling page as well:
|
In practice, oversampling is implemented in order to achieve cheaper higher-resolution A/D and D/A conversion. For instance, to implement a 24-bit converter, it is sufficient to use a 20-bit converter that can run at 256 times the target sampling rate. Averaging a group of 256 consecutive 20-bit samples adds 4 bits to the resolution of the average, producing a single sample with 24-bit resolution. Note that this averaging is possible only if the signal contains perfect equally distributed noise (i.e. if the A/D is perfect and the signal's deviation from an A/D result step lies below the threshold, the conversion result will be as inaccurate as if it had been measured by the low-resolution core A/D and the oversampling benefits will not take effect). |
I don't understand how averaging a group of 256 consecutive 20-bit samples adds 4 bits to the resolution average. 
My mind is drifting... |
|
g
Member
Topics: 10
Replies: 2037
Registered: 03.Jan.03 |
Atlantis wrote on 11 Dec. (8:14) :
I don't understand how averaging a group of 256 consecutive 20-bit samples adds 4 bits to the resolution average.
|
Let's say you have a 4 bit sample, which means it can contain 16 discrete values from 0 and 1, and you want to play it back using a 2 bit D/A. The problem is, the 2 bit D/A only has four discrete values, 0, 1/3, 2/3 and 1. If you want to use overampling to get a higher average bit depth, you have put out your bits faster. Twice the rate won't do, it gives only 7 discrete averages. Three times the rate gives 10 discrete values and four times 13. Five times the rate finally gives us 16 different averages, exactly what we need to reproduce our 4 bit sample. It works the same way with 24 bit samples on 20 bit D/A:s, only the numbers are too many to keep in my head. 
klart slut |
|
Ethelion
Member
Topics: 40
Replies: 562
Registered: 25.Mar.03 |
Yes, I can.
48khz is the only solution for me.
The difference is mindblowing once you get a pair of good speakers/monitors, or a quality headset.
I can play World of Warcraft using 41khz, but that's about it.
| My music : CTGMusic 
| My art : DeviantArt 
| My life : ethelion.com  |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
I decided to do a new one, this time processing the first one at 192,000 Hz, and resampling down to 44,100 Hz with an anti-alias filter, and then adding the second one at 44,100 Hz, and dithering the output down to 16 bit. Which one sounds clearer, and can you hear the buzzing?
http://atlanteanrecords.googlepages.com/anti-aliasing.rar
Here are the two spectrums (prior to dithering) (click to enlarge):
My mind is drifting... |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
|
s-f wrote on 11 Dec. (10:50) :
Atlantis wrote on 11 Dec. (8:14) :
I don't understand how averaging a group of 256 consecutive 20-bit samples adds 4 bits to the resolution average.
|
Let's say you have a 4 bit sample, which means it can contain 16 discrete values from 0 and 1, and you want to play it back using a 2 bit D/A. The problem is, the 2 bit D/A only has four discrete values, 0, 1/3, 2/3 and 1. If you want to use overampling to get a higher average bit depth, you have put out your bits faster. Twice the rate won't do, it gives only 7 discrete averages. Three times the rate gives 10 discrete values and four times 13. Five times the rate finally gives us 16 different averages, exactly what we need to reproduce our 4 bit sample. It works the same way with 24 bit samples on 20 bit D/A:s, only the numbers are too many to keep in my head.
|
Now that's a lot of smart stuff.  How does twice the rate give 7 discrete averages, etc.? I kind of see where you're coming from but I don't think I understand it fully.
My mind is drifting... |
|
LPChip
 Administrator
Topics: 39
Replies: 1261
Registered: 05.Jun.05 |
If you use plugins that apply somekind of filter, and its buggy, you can actually hear a difference between 44.1 and 48khz. If not, then it should be impossible. I used to think that I could hear the difference between 48 and 96, but that appeared to be caused by buggy plugins. After complaining about it and getting a good version, it was equal to me. I'm on a X-Fi Elite Pro card so that can also be the reason. Its very HQ 
You can ask me hundreds "why don't you" questions, answer will be, because it works that way too. - xTr1m |
|
oZostomia
 Artist
Topics: 48
Replies: 506
Registered: 23.May.04 |
It´s all about the finest detail you can get out of music!
And now i dont talk about trance or metal music...
Little like Norah Jones, Samuli Edelmann type of music that need to get the most
finest deails in good quality so the listener can enjoy the music!
But the listener needs pretty good pair of speakers too to get the most fun out of the music listening!
"....How can you hear what is right or wrong?...."
 |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
LPChip wrote on 12 Dec. (15:07) :
If you use plugins that apply somekind of filter, and its buggy, you can actually hear a difference between 44.1 and 48khz. If not, then it should be impossible. I used to think that I could hear the difference between 48 and 96, but that appeared to be caused by buggy plugins. After complaining about it and getting a good version, it was equal to me. I'm on a X-Fi Elite Pro card so that can also be the reason. Its very HQ
|
What's with everyone saying it's impossible? The samples are right here, and I can hear the difference easily, even more so with my AKG K 240 S, and I'm using a MOTU 828mkII FireWire. Maybe start here if you don't know what to listen for, and turn the volume up because it's low-level distortion.
My mind is drifting... |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
Oh oops, I should've used a sawtooth waveform.  Doh! Let me do them again. 
My mind is drifting... |
|
|
Latest Topics
![[203.118.142.209]](pics/ip.gif){kind=small}
