Digital Sound

Objectives
• Properties of Digital Sound
• Digital Sampling of Sound
• Introduction to MIDI
• Synthesizing Audio
• Speech Synthesis and Recognition

cabangpokok
 Properties of Digital Sound
• Definition
– Digital sound is a sound that has been converted
to or created in a discrete form suitable for
storage and processing in a computer.
• Once in digital form, sound can be modified or
edited in a variety of ways.

cabangpokok
 Properties of Digital Sound
• Sound is produced by the vibration of some
membrane.
• These vibrations are then transmitted as
waves through a medium:
– air
– water
– some other

cabangpokok
 Properties of Digital Sound
• To measure the intensity of a tone, it would be
plotted as a wave.

cabangpokok
 Properties of Digital Sound
• When the analog wave meets our ears, it
causes our eardrums to vibrate and transmit
the sound as a signal to our brain.
• Recording and playing digital sound require
the conversion of the analog signal to discrete
form.

cabangpokok
 Frequency
• Length of time it takes the wave to complete
one entire cycle Heinrich Rudolf
Hertz
• Unit = Hz (Hertz), cycle per second
• Determines the pitch of the sound as heard by
our ears
– The higher the frequency, the higher the pitch of
sound

cabangpokok
 Frequency

Quiz:
• f = 4.0, T=?
• f = ?, T=3
cabangpokok
 Frequency
• Infrasonic
– 0 to 20 Hz
Seismic
– Lower than ‘normal’ limit
• Audiosonic
– 20 Hz to 20 kHz
– Human hearing range
• Ultrasonic
– 20 kHz to 1 GHz
– Higher than human hearing
• Hypersonic
– 1 GHz to 10Thz
– Exceeding speed of sound
– Hypersonic is one of three rough classifications for airflow regimes
• subsonic (less than Mach 0.8),
• transonic (Mach 0.8 - 1.2),
• supersonic (Mach 1.2 - 5.0), Sonic Boom
• and finally hypersonic (Mach 5.0+)

cabangpokok
 Amplitude
• Sound’s intensity or loudness
• Pitch refers to frequency
• Loudness refers to amplitude and
• Timbre refers to harmonic content of the
sound

cabangpokok
 Digital Sampling of Sound
• Sampling a sound wave consists of
determining the amplitude of the sound at
some number of discrete times within the
interval
• Generally, these discrete times equal distances
along the time axis
• The number of these times chosen per second
determines the sampling rate (in Hz)

cabangpokok
 Digital Sampling of Sound
• A sampling rate of 22
Hz means:
– Dividing the time axis
into 22,000 equally
spaced times per
second
• Choice of sample rate
is a function of the
capability of ADC
(Analog-Digital
Converter)
cabangpokok
Digital Sampling of Sound

cabangpokok
 Nyquist’s Theorem
• “For lossless digitization, the sampling rate
should at least twice the maximum frequency
response”
• If sampling rate is 44100Hz, the maximum
frequency could be capture is 22050 Hz

cabangpokok
 Nyquist’s Theorem
• Careful with low samples per cycle
• If too low, upon reconstruction the analog
wave that the sound had a constant amplitude
(monotonous sound)
• This phenomenon called Aliasing

cabangpokok
 Nyquist’s Theorem
• Careful with low samples per cycle
• If too low, upon reconstruction the analog
wave that the sound had a constant amplitude
(monotonous sound)
• This phenomenon called Aliasing

cabangpokok
 Return: Digital Sampling of Sound
• The quantization process consists of
converting a sampled signal that can only have
a limited number of values.
• 8-bit quantization provides 256 possible
values
• 16-bit quantization results more than 65,536
possible values

cabangpokok
 Audio Quality
• Several factors contribute:
– Quality of audio source
– Quality of hardware used to capture audio
(functionality etc..)
– Properties used during capturing process
• Frequency
• Sampling rate
• No of channel
– Mono or stereo

cabangpokok
 Audio Size
• Size of an audio file depends on
– Length in seconds (t)
– Sampling rate in Hz (S)
• Number of sample obtained in 1 second
– Resolution bit in bit/byte (B)
– Number of channel used (C)
• Mono or stereo

• Size of Audio File = t x S x B x C

cabangpokok
 Audio Size
• Size of Audio File = t x S x B x C
• Example
– 30 seconds mono audio with sampling rate of
10Hz and resolution of 1 byte.
• ??
– 1.5 minute stereo audio with sampling rate of
20Hz and resolution of 8 bit.
• ??

cabangpokok
 MIDI
• Musical Instrument Digital Interface
• Very efficient method of representing musical
performance information
• MIDI data file are extremely small when
compared with audio files
– 10 minutes sampled audio – 10 Mbytes/minute
– 10 minutes of typical MIDI – less than 10
Kbytes/minute
• MIDI protocol provides a standardized and
efficient means of conveying musical
performance information as electronic data
cabangpokok
 MIDI
• MIDI interface is composed of two different
component:
– Hardware to connect physical equipment
• Adds a midi port to an instrument, specifies a MIDI cable and
processes electrical signal receives over the cable
– A data format that encodes information to be
processed by the hardware
• Specific data format for each instrument
• Describing
– start and end of scores,
– basic frequency,
– Loudness
– Instrument itself
– Csound Demo
cabangpokok
 Synthesizing Audio
• One of the earliest uses of computer for audio
processing
– Sound synthesis systems
• Used to produce music created with electronic
instruments
• Oscillator as a sound source
• Electronic oscillator
– electronic circuit that produce repetitive
electronic signal

cabangpokok
 Simple Waveforms
• Oscillators in a synthesis system produce very
simple waveforms
• 4 basic types
– Sine
– Square
– Triangle
– Sawtooth

cabangpokok
cabangpokok
 Basic Synthesis Technique
• Subtractive Synthesis
– Early method and still used for analog
synthesizers
– Start with very complex waveform
• Signals from several oscillators + specific types of noise
• White noise
– Noise produced by combining sounds of all different
frequencies together (all human can hear)
– Combination of all different frequencies
– Filters are then used to remove unwanted noise
– Leaving desired sound
cabangpokok
 Basic Synthesis Technique
• Additive Synthesis
– Start with simple waveforms and building more
complex waveforms
– Allows mathematical computation of component
sine waves that can be combined to forma given
complex wave

cabangpokok
 Synthesis Technique
• Frequency Modulation (FM)
– Two simple waveforms
– One wave (modulation) is used to
modify the other (carrier) to a more
complex form
– Produce rich sound, but hard to
control

cabangpokok
 Synthesis Technique
• Phase Distortion Synthesis
– Distort a simple waveform by modifying the time
scale at different rates, changing the time it takes
for a portion of a cycle to be completed

cabangpokok
 Synthesis Technique
• Integrated Synthesis or Variable Architecture
Synthesis
– Many waveforms have similar basic shape or
envelope
– Referred as an ADSR (attack, decay, sustain,
release) form

cabangpokok
 Synthesis Technique
• ADSR
– Many waveforms have similar basic shape or
envelope
– Referred as an ADSR (attack, decay, sustain,
release) form

cabangpokok
 Speech Synthesis and Recognition
• Artificial Intelligent of Human Speech
• Most promising use of computer’s sound
capabilities is computer/human interaction
using spoken language
• Ex: Stephen Hawking
• How it works?

cabangpokok
 Speech Synthesis and Recognition
• Computer Speech Synthesis
– Generating speech from written text
– How? Store the pronunciation
– Problems:
• Large storage
• Slang words
• Difficult to achieve natural speech rate
• Same spelling, but different in context (ex. read)

cabangpokok
 Speech Synthesis and Recognition
• Computer Speech Synthesis
– Alternate for speech synthesis – analysis the
written text before trying to pronounce it
– Brake the text selection into sequence of basic
phonemes

cabangpokok
 Automated Speech Recognition
• Digitized the speech, then reconstruct the
words were spoken
• Human voice – fingerprint
• Voice recognition system – specific person
• “Speaker-dependant speech recognition
system”

cabangpokok
cabangpokok
 Automated Speech Recognition
• Digitized the speech, then reconstruct the
words were spoken
• Human voice – fingerprint
• Voice recognition system – specific person
• “Speaker-dependant speech recognition
system”
• Voice activation system
• SIRI
cabangpokok