Intro to digital audio

Question 1

How does digital audio differ to analogue?

Answer 1

Represents sound as a sequence of discrete numbers, rather than a continuous variation of pressure / voltage

Question 2

Why is digital audio easier to transmit, store and copy reliably?

Answer 2

Has a fixed temporal (and thus frequency) resolution
 and fixed amplitude resolution. Whereas analogue has a theoretically infinite temporal and amplitude resolution.

Question 3

When was the first computer based digital recording system made?

Answer 3

1957 by Max Matthews

Question 4

When was the first CD player made?

Answer 4

1982 - Sony CDP-101

Question 5

When was the ADAT made?

Answer 5

1991 - used video tape to store 8 tracks of audio.

Question 6

What do more bits mean?

Answer 6

More bits mean more values representable. Adding a bit doubles the range of possible values.

Question 7

What is increasing bit depth used for?

Answer 7

• To record audio samples with greater precision / resolution.
• Because each bit doubles the range of amplitudes,
it adds 6dB to the signal-to-noise ratio (SNR): 

e.g. 16-bit = 16 x 6 = 96dB

Question 8

What are individual bits transmitted as?

Answer 8

• fixed high and low voltage levels (e.g. S/PDIF +/- 0.3V)

- on/off flashes of light (optical)

Question 9

What is the sampling theorem?

Answer 9

• Provided certain limitations are observed, any arbitrary wave form can be recorded by taking individual samples of amplitude at fixed intervals.
• The wave form is played back by outputting the same samples at the same fixed intervals.

Question 10

What is the sampling frequency?

Answer 10

The number of samples taken over the course of one second:

• measured in Hz
• known as the sample rate

Question 11

What are different sample rates used for?

Answer 11

Different purposes based on different needs (quality, bandwidth).

Question 12

What is the sample rate of telephony?

Answer 12

8kHz

Question 13

What is the sample rate of broadcast (eg. radio)?

Answer 13

32kHz

Question 14

What is the sample rate of a CD?

Answer 14

44.1kHz

Question 15

What is the sample rate of audio for video?

Answer 15

48kHz

Question 16

What is the sample rate of a professional recording?

Answer 16

96kHz

Question 17

When does aliasing occur?

Answer 17

When the input frequency is greater than half the sampling frequency. This point is called the Nyquist limit.

Question 18

Describe the frequencies above and below the Nyquist limit (fN).

Answer 18

Frequency components above fN are mapped
 to (and heard as) frequencies below fN. Thus, all frequencies in the input signal 
must be less than half the sampling rate

Question 19

What happens to frequencies above the Nyquist limit?

Answer 19

Frequencies above the Nyquist limit are mirrored around it: At a 48kHz sample rate: an ordinarily inaudible frequency of 10kHz above Nyquist (34kHz) will be audible as 10kHz below Nyquist (14kHz).

Question 20

How are frequencies above the Nyquist removed?

Answer 20

This is done with a low-pass filter known as an analogue antialiasing filter.

Question 21

What is oversampling used for?

Answer 21

To avoid the use of analogue brick-wall filters (which exhibit phase problems). Instead, an analogue gentle-slope filter and a digital brick-wall filter are combined

Question 22

What are the advantages and disadvantages of a brick wall filter when oversampling?

Answer 22

• Good frequency response

- Poor phase response

Question 23

What are the advantages and disadvantages of a gentle filter when oversampling?

Answer 23

• Poor frequency response (higher frequencies attenuated)

- Good phase response

Question 24

How do you avoid the gentle slope filter from affecting audible frequencies?

Answer 24

Oversample the audio at a much higher rate than required (e.g. from 4x to 128x).

Question 25

How can a digital filter help with phase shifting?

Answer 25

Digital filters can have linear phase response, so the subsequent brick wall filter does not cause a phase shift.

Question 26

What happens to the oversampled audio?

Answer 26

The oversampled audio is digitally down-sampled to the required sample rate.

Question 27

What are band limited synthesisers used for?

Answer 27

In digital synthesisers, band-limited versions 
are used to ensure the harmonics do not
 exceed the Nyquist frequency.

Question 28

What happens to the amplitudes after sampling?

Answer 28

After sampling, the amplitudes are stored as 16-bit (CD) or 24-bit (DVD) integer values.

Question 29

When is the best SNR (signal to noise ratio) achieved?

Answer 29

The best SNR is achieved when signal spans the full range of levels.
- CD: 16-bit (96dB), DVD: 24-bit (144dB).
Weaker signals use fewer levels (bits) and so have a worse SNR.

Question 30

What does dithering do?

Answer 30

Adds a small amount of random noise to de-emphasise the steps
- Digital ‘blurring’
to hide the lines

Question 31

What does dithering improve?

Answer 31

Improves playback
 sound perceptually

Ensures noise is
 broadband, which 
is harder to notice (
not correlated 
with the audio)

Question 32

What else will distort audio?

Answer 32

Uneven / irregular clock pulses will also distort the audio.

• Leads to phase issues, spectral distortion, poor stereo imaging.
• Fundamentally a hardware limitation 
(not easily solved in software)