Sampling Rates....PLEASE HELP

[danbs10]

Hi, please can someone translate this into lamens English for me:

Quote taken from Wikipedia:

The Nyquist–Shannon sampling theorem states that perfect reconstruction of a signal is possible when the sampling frequency is greater than twice the maximum frequency of the signal being sampled, or equivalently, when the Nyquist frequency (half the sample rate) exceeds the highest frequency of the signal being sampled. If lower sampling rates are used, the original signal's information may not be completely recoverable from the sampled signal.[2] For example, if a signal has an upper band limit of 100 Hz, a sampling frequency greater than 200 Hz will avoid aliasing and allow theoretically perfect reconstruction.

From this please tell me what I should set my input and output sample rates at.


Many thanks

 

[Emerson Maningo]

Audible sound frequency is from 20Hz to around 22,000Hz. According to the theory: "sampling frequency is greater than twice the maximum frequency of the signal being sampled". So the sampling rate for acceptable reconstruction of audio signal will be 2 x 22,000Hz or around 44.1KHz.

It is why in digital audio particularly the red book CD, sampling rate is around 44.1KHz. In recording, you might want to set more than that for more accurate reconstruction, for example the common sampling rate used in recording would be 48KHz. And in some projects, they may even use higher sampling rate such as 96KHz and 192KHz although that is rare because the larger the sampling rate, the bigger will be the resulting file size of your audio.

 

[Bobbsy]

There's no single correct answer to this. Some believe that, despite Nyquist, they can hear the difference in quality as sample rates go up (and this maybe true when effects are added to the equation).

However, to start off, I suggest you just use 44.1kHz. It'll put less pressure on your computer hardware and avoid sample rate conversion when you decide to burn to CD. Later on, if you wish, experiment with higher sample rates and see if any perceived quality gain is worth the extra disk storage and if your system can handle all the extra bits.

...and input and output really need to match.

 

[DD83]

The sample rate is basically like a camera. When you sample something it is like taking a picture of a moving car multiple times. looking back at the photos there are going to be gaps where the vehicle moved that you didn't capture. The higher the sample rate the more "photos" taken. So let's say at 8 bit you were able to get 50 photos, at 12 bit you got 75 and 16 you got 100. That's essentially what the sampling rate is/does. The higher the rate the more pictures taken. So in theory the highest sampling rate is going to give you the most accurate representation of what you recorded. I still use 16 because CD quality is better that mp3 quality that most people listen to now a days. it does save space as well.

 

[Farview]

OK, clearing some stuff up in this thread:

It is why in digital audio particularly the red book CD, sampling rate is around 44.1KHz. In recording, you might want to set more than that for more accurate reconstruction, for example the common sampling rate used in recording would be 48KHz.

A higher sample rate will not give you more accurate reconstruction, it will just allow you to record and reconstruct higher frequency material. If you are not recording anything above 20k, there is no difference between recording at 44.1k and 96k. (however, some converters might sound better at one sample rate than the other, but that is a function of design implementation and not the sample rate itself)

48k sample rate is used when recording audio for video because, mathematically, it will fit better with the video frame rates and won't have to be resampled to avoid audio drift.

The sample rate is basically like a camera. When you sample something it is like taking a picture of a moving car multiple times. looking back at the photos there are going to be gaps where the vehicle moved that you didn't capture. The higher the sample rate the more "photos" taken. So let's say at 8 bit you were able to get 50 photos, at 12 bit you got 75 and 16 you got 100. That's essentially what the sampling rate is/does. The higher the rate the more pictures taken. So in theory the highest sampling rate is going to give you the most accurate representation of what you recorded. I still use 16 because CD quality is better that mp3 quality that most people listen to now a days. it does save space as well.

ummmm.....no. None of this post is accurate and some of it doesn't even make sense. Sorry.

To answer the question: 44.1k is for audio for CD. 48k is for audio for video. 96k is for when it is necessary to capture audio above 20kHz. (which really isn't often)

 

[Emerson Maningo]

Just some clarification. Definition of sampling rate(from Wikipedia): "The sampling rate, sample rate, or sampling frequency (fs) defines the number of samples per unit of time (usually seconds) taken from a continuous signal to make a discrete signal."

So if it is the number of samples per unit time taken from an analog(continuous) signal. Wouldn't sampling at 96,000 samples per second (96KHz) more accurate than 48,000 samples per second (48KHz) or even 44100 samples per second (44.1KHz)?

Why other professional recording equipment or standards do use sampling rates above 44.1KHz, as I checked it here: Sampling rate - Wikipedia, the free encyclopedia

Are they doing it for accurate reconstruction or for what?

 

[Farview]

Just some clarification. Definition of sampling rate(from Wikipedia): "The sampling rate, sample rate, or sampling frequency (fs) defines the number of samples per unit of time (usually seconds) taken from a continuous signal to make a discrete signal."

Correct.

So if it is the number of samples per unit time taken from an analog(continuous) signal. Wouldn't sampling at 96,000 samples per second (96KHz) more accurate than 48,000 samples per second (48KHz) or even 44100 samples per second (44.1KHz)?

That depends on what you are trying to record. The only thing a higher sample rate allows you to do is record higher frequencies. If there are no higher frequencies in the signal that you are trying to record, then you can have all the accuracy there is to get with the lower sample rate.

For example, if you have a 10khz sine wave that you are trying to record, you can record it and reconstruct it easily with a 44.1k sample rate. Recording it at 96k will not get you a more accurate recording of the 10kHz sine wave because 44.1k can already reconstruct the whole thing. There is no more detail to record or reproduce, in fact 44.1k is already twice the sample rate you need to reproduce all the detail that a 10kHz sine wave possesses.

So, since most people can not hear anything above 20kHz, most microphones to pick up much above 20kHz, most instruments don't give off much sound above 20kHz, and most stereos won't play back anything above 20kHz, you don't need a sample rate higher than 44.1k to reproduce all of the information that you are trying to capture.

Why other professional recording equipment or standards do use sampling rates above 44.1KHz, as I checked it here: Sampling rate - Wikipedia, the free encyclopedia

Are they doing it for accurate reconstruction or for what?

They are doing it primarily because they believe that there are frequencies that we can't hear, but still affect the frequencies that we can hear. That's a debatable point. But realistically, if it made an obvious difference, there would be no debate. No one debates whether it's better to use a spoon or a shovel to dig a large hole, because it's obviously better to use a shovel.

Some of the super high sample rates are are DSD audio instead of PCM, they work on a completely different premise since they are only 1 bit. So they can't be compared in this discussion.

Other odd sample rates were only used on one machine before any standards were put in place, so they don't count either.

 

[TimOBrien]

Easy answer for the REAL WORLD:

If you're doing video, use 48k (video needs 16bit/48k final files to sync to video)
For music use 44.1k

Record at 24bit/44.1k and render down to 16bit/44.1k

Unless you have a mega-dollar studio, forget anything else....

 

[Bobbsy]

Let me start by saying that I'm firmly in the "use 44.1" camp and recommend that for the OP as well. (I use 32 bit float rather than 24 bit integer then convert to 16 bit...but that's another discussion.)

However, the argument about using higher sample rate is more than just "there are other frequencies that affect the sound". There's also a thought that, when you use software based effects, particularly things like pitch shifting, the higher sample rate can come into play.

The other thing that can affect the sound is not the sample rate itself but the quality of the filters used in the A to D converters. Aliasing can be an issue and this can be reduced through the use of higher sample rates (not because of the sample rate but because of the different filter frequencies that the change entails).

Anyhow, I hope this doesn't confuse the OP but I find the discussion interesting. In the real world (at least unless you have a multi million dollar studio) I agree with the advice to use 44.1 for normal audio and 48 for video-related work.

 

[Farview]

Let me start by saying that I'm firmly in the "use 44.1" camp and recommend that for the OP as well. (I use 32 bit float rather than 24 bit integer then convert to 16 bit...but that's another discussion.)

However, the argument about using higher sample rate is more than just "there are other frequencies that affect the sound". There's also a thought that, when you use software based effects, particularly things like pitch shifting, the higher sample rate can come into play.

The other thing that can affect the sound is not the sample rate itself but the quality of the filters used in the A to D converters. Aliasing can be an issue and this can be reduced through the use of higher sample rates (not because of the sample rate but because of the different filter frequencies that the change entails).

All of this is true, but the difference is extremely subtle.

The point I was trying to make is that any sound difference heard between 44.1k and higher sample rates like 96k is a difference in the design and implementation of the converters you are using and not a function of the sample rate itself. This is one of the reasons why you get so many people swearing that one is better than the other. On the set of converters that they are using, one sample rate could sound better than another. And the guy he is arguing with could be using a different set of converters that sound better at a different sample rate than the first guys converters do.

As far as plugins go, that is a case by case basis. Some of the better plugins actually upsample the audio before it processes it and then downsamples the audio back into the session, so it doesn't matter what sample rate the session is.