Subject:Best sample rates?
Posted by: Bush
Date:11/6/2002 11:50:42 PM
What’s the difference between 48,000 96,000 192,000 sample rates and what do you recommend that a person use for the best sound for streaming and for the best sound on a CD that I will sell? Thanks, Gus |
Subject:RE: Best sample rates?
Reply by: jasonb
Date:11/7/2002 2:33:07 AM
most pros record at 24 bits/96000 hz or 24 bits/48000 hz then later dither the file back down to 16 bits/44100 for cd burning. in theory higher bits/hz will produce more better recording, but it also increases file sizes, so you'll have to experiment and adjust for size vs quality. I personally use 24bits/48000hz just my 2 cents |
Subject:RE: Best sample rates?
Reply by: ramallo
Date:11/7/2002 6:35:08 AM
Hello, With highest sampler frecuencys (PCM), you optain a better work of the low-pass filter (more far from audio) and better impulse response. In my particular case, I can't ear any diference between a take of 96, and the same take resamplig at 48. I use 24/48 Cheers |
Subject:RE: Best sample rates?
Reply by: drbam
Date:11/7/2002 8:16:41 AM
<<In my particular case, I can't ear any diference between a take of 96, and the same take resamplig at 48. I use 24/48>> So then, what is the audio quality advantage of 24/48 over 24/44.1? drbam |
Subject:RE: Best sample rates?
Reply by: Rednroll
Date:11/7/2002 9:43:23 AM
Here's some posts from the past that I described sampling frequency and bit depths and some other discussions on this subject. You might have to wade through some mud in the strings, but there's some good info there. http://www.sonicfoundry.com/forums/ShowMessage.asp?ForumID=4&MessageID=62577 http://www.sonicfoundry.com/forums/ShowMessage.asp?ForumID=19&MessageID=100863 http://www.sonicfoundry.com/forums/ShowMessage.asp?ForumID=3&MessageID=58744 |
Subject:RE: Best sample rates?
Reply by: Rednroll
Date:11/7/2002 12:05:11 PM
"So then, what is the audio quality advantage of 24/48 over 24/44.1?" The difference can be easily understood by knowing about the "Nyquist Theorm". The Nyquist theorem basically says to accurately reproduce a particular frequency the sampling frequency must be twice that particular frequency. So for a 44.1khz sampling frequency the highest audio frequency that can be reproduced would be 22,500 Hz. For a 48Khz sampling frequency the highest audio frequency would be 24Khz. So that's the difference. Human hearing is limited to a bandwidth of 20Hz to 20Khz. I've seen a study in college performed which showed most people could not detect a frequency over 15Khz. So if you can tell me what 23Khz or 24Khz sounds like, then please describe it to me. Most of the time the argument is that these frequencies can be "felt" and not heard. Well if that is true, then I wonder why they limit frequencies in the color spectrum, because I can then come with the argument that these colors are "felt" and not seen. So for those "pros" recording at 96Khz to be able to hear frequencies at 48Khz I'm one pro who's not buying it, I'll save my money and invest in things that can actually be heard. |
Subject:RE: Best sample rates?
Reply by: Sonic
Date:11/7/2002 3:21:48 PM
As the previous threads posted by Red probably state, it isn't about whether you can hear > 22 kHz. Yeah, there's some research about bone conduction and whatnot, but the *real* reason to use higher sampling rates is exactly what ramallo was trying to say. If you have an extra 20 kHz of inaudible bandwidth below Nyquist, the requirements of the anti-alias filter in the A/D are drastically reduced and you don't have to use steep, phase-altering filters (which *can* produce pretty audible results). Similarly, the anti-image filter in the D/A needn't be as stringent to more transparently recover the audible range. But that does depend on being able to tweak your i/o hardware (or the hardware being fairly intelligent with oversampling, etc.). Even if you have the hardware, doubling disk and processing requirements is a serious issue to factor in. Try recording the same piece of (live) material at 96 kHz and 48 kHz and see if you can tell the difference (CD's don't really count since they are already bandlimited). J. |
Subject:RE: Best sample rates?
Reply by: Rednroll
Date:11/7/2002 5:22:12 PM
Yes, this is totally correct when you are recording and having to deal with an Analog Low-pass circuit. This is where the term "over sampling" comes into play though. Although you may be recording at 48Khz, the hardware is usually sampling 4 times that when going from A/D. The hardware is designed to put a low-pass filter with a cut-off frequency of 4 times your sampling frequency when the signal is analog, thus allowing a lot of higher order frequencies into the A/D converter and not introducing phase shift due to the analog filter as Sonic mentioned. Then once it is converted to Digital just after this is another low-pass filter. Once you're in the digital stage you can put a "Brickwall" filter without having any phase effects, so now what is at a 192Khz sampling frequency (4 x 48K) goes through a digital brick wall low-pass filter set at 24Khz and then the audio is resampled back down to the 48Khz sampling frequency. You can not use a brickwall filter when you're in the analog world, because one it is not possible and 2 it introduces a bunch of phase shift problems as previously mentioned by Sonic. Most newer hardware is set up this way and the user does not have to worry about it, so recording at 96Khz really does you no good, unless you enjoy using up twice the amount of hard drive space. |
Subject:RE: Best sample rates?
Reply by: Geoff_Wood
Date:11/8/2002 2:11:16 AM
Anything that you are going to be streaming with a view to people listening on computer speakers should be just fine at 22KHz / 16 bit IMO. WRT hearing frequencies, anybody over 15 years is going to be hard-pressed to hear past 17KHz. Some say that we can however discern the effects of frequencies over 20KHz, though there is debate about this even at the highest level of experts, so it must be *very* subtle. And listening with what speakers ?!!! Check out samples at www.pcabx.com to see if you can hear *any* difference between the higher bit depths and sample rates in a proper testing environment, if you have hardware that can pass it ! geoff |