View Full Version : 24 bit recording?

11-30-2009, 09:57 PM
I've been looking at small digital recorders, and some record in 24 bit.

What does that give you that 16 bit doesn't?

I understand that 48 khz would give you higher frequencies than 44.1 or 22 or 11, but what does a higher bit depth give you?

And, since it's higher than CD quality, can the difference be heard? Do MP3's sound better if made from 24 bit?

11-30-2009, 11:12 PM
The sample rate is the number of samples per unit of time, in this case, per second. . . analogous to the frame rate in a video. 44.1k is the standard CD sample rate.

The bit depth is the number of bits per sample, analogous to the resolution in pixels of any given frame in the above video. Greater bit depth is better resolution.

so. . . you have two gradients affecting the total amount of data processed per unit of time.

Unless you have a special application in mind, record at 24 bit, 44.1k sample rate. It'll give you max headroom, you won't have to mess with conflicting sample rates which can cause problems, and you can dither it down to 16 bit if you want to put it on a CD.

I'm still pretty new to home recording, so someone here may be able to give you better information. AFAIK, For the purpose of compression, generally a good rule of thumb is the better the source material, the better the end product. Once it's compressed, you probably won't be able to tell if the source material was 16 or 24 bit.

12-01-2009, 12:22 AM
Scottie pretty much covered it.

I went looking for a picture of the difference between more bits and less and found this nice explanation of digital audio in general: CLICKY (http://www.joelstrait.com/blog/2009/10/12/a_digital_audio_primer)

One of the things you want to do with any recording (digital or analog) is make sure the levels on each track are as high as possible without causing clipping (http://en.wikipedia.org/wiki/Clipping_%28audio%29). This lets you take full advantage of the dynamic range you get out of your medium. To put that another way, it's better to have to cut the level of a recorded signal back by reducing dynamic range than it is to try and boost it by adding information that isn't there to begin with.