We looked previously at the sampling theorem and the way that affects the frequencies which can be represented. Since the advent of audio compression algorithms all the talk has been about bit rate. Today we’ll look at these two terms and converting between the two.

As we’ve said previously, the sample rate is simply the number of times we record the signal’s value when we convert it from analogue to digital. There are more values that become important at this point. The nature of a digital audio system is that it will only allow a sample to take a finite number of values. The number of bits (a bit is a binary one or zero) a sample is recorded with is called the sample resolution. In CD audio the sample rate is 16bits. The other thing to remember about CD audio is it is stereo.

So to get to our bit rate we multiply:

Sample Rate x Sample Resolution x Channels

For CD audio that’s:

44100 x 16 x 2 = 1411200 bits per second or 1.4 Megabits per second.

Seeing this figure makes those of us familiar with compressed audio realise the sort of data rate savings it brings. “High quality” compressed audio tends to have a bit rate of between 300-400 kilobits per second.

### Like this:

Like Loading...

Of course it is true that 44.1 kHz * 16-bits * 2 channels is used for mastered CDs but no-one professionally records music at this rate anymore. If we were to use Raspberry Pi to implement a silent audio recording system (a Digital Audio Workstation) the we must consider professional rates. Starting with a 96 kHz by 16-bit channel we have a bit rate of 1.536 Mbps per channel. Therefore, an 8 channel recorder needs a total communications bit rate of 12.288 Mbps, and if you add for return channels for monitors/headphones this increases to 18.432 Mbps.

Of course 96 kHz is the starting rate for professional systems; the better ones use twice this rate: 192 kHz, which would also double all the bit rates stated above, although for a Raspberry PI DAW I would suggest using 96 kHz.