Lecture 2: Digital Sound Processing

Digital Sound Processing

Digital Sound Processing System in Stages:

• Signal In
• Band Limiting
• Analogue to Digital Conversion
• Digital Signal Processing Operations
• Digital to Analogue Conversion
• Smoothing
• Signal Out

Signal Processing is an area of electrical and systems engineering along with an applied mathematics formula which analyses and performs operations on signals in either discrete or continuous time periods in order to produce useful operations from these signals.

Signals are analog or digital and electrically represent a variation in time or space in physical quantities.

Electronic Filters

Electronic filters are electronic circuits which perform functions of signal processing. There can be two reasons for this

• Remove frequency components from the signal
• Enhance existing frequency component

Electronic filters can be:

• Passive (a component that consumes but does not produce energy or a component incapable of gaining power) or Active (a type of analogue electrical filter, recognised by the use of one plus activite components, such as different types of amplifiers, such as voltage or buffer
• Analogue or digital
• High Pass Filter- a device that passes high frequencies and reduces the amplitude of frequencies which are measured at more than its maximum cut off point
• Low pass filter- passes low frequency signals but reduces the amplitude of signals with above maximum frequencies, it is the opposite of a high pass filter and is also known as a high-cut filter or a treble cut filter when used in audio applications
• A band pass filter- a combination of high and low pass filters.
• A band stop filter/band rejection filter- passes most frequencies in their original state but alters some within a specific range
• Discrete time or continous time
• Linear or non Linear
• Infinite Impulse Response- deals with filters with an impulse response over an infinite length of time or Finite Impulse Response which give fixed time responses

Pitch

Pitch is a perceptual concept with allows the human ear to oder sounds on a frequency related scale. High and low pitches are compared in relation to musical melodies, which require sound with a claer frequency is clear and stable enough to be heard as something more substansial than just plain noise. Pitch is considerd a major auditory attribute of musical tones, along with duration, loudness and timbre.

Digital Signal Processing System Requirements:

• Input and output filetring
• Conversion from analogue to digital and vice versa
• Digital processing unit

Why choose Digital Signal Processing?

1. Precision

Although theoretically digital signal processing is limited only by the process converting input and output (analogue to digital and digital to analogue) in practice word length (number of bits) and sampling rate (sampling frequency) aid modifications. The ever increasing operating speed and modern word length is increasing the areas of application.

2. Robustness

Logic level noise margins benefit digital systems, making them less susceptible to electrical noise and component tolerance variations, in comparison to analogue systems. Vitally, adjustments in complex systems for electrical drift and component ageing are virtually removed in complex systems.

Electrical noise is a random fluctuation in an electrical signal. There is a large scope for noise generated by electronic devices as there are various different effects which can lead to it being produced.

Component tolerance variations may be specified as a number of things:

• Factor/Percentage from the nominal value
• Maxiumum deviation from the nominal value
• Explicit range of allowed values
• Implied by the numeric accuracy of the nominal value

Electrical drift is when an undesired progressive frequency change occurs. Main causes for this are the ageing of the components and changes in the environment. It can happen in either direction, the frequency can increase or decrease.

3. Flexibility

The programmability of digital signal processing caters to aid processing operations by expansion and upgrading, without majorly significant hardware changes. It is possible for a user to constuct a practical system with suitable characteristics such as time varying in order to allow adaptations.

Sound Card Architecture

• Spatial anti-aliasing is the technique that helps minimise aliasing when representing a high-resolution image at a lower resolution than its original state. This means that if you are attempting to reduce a pictures file size and display it, the detrimenta, l effects of the picture will be as limited as possible. It is used in many applications, including digital photography and computer graphics. Anti-aliasing is often used prior to converting from analogue to digital in order to remove the out of band component of the signal

Sampled Data Reconstruction Filters

• The input of Analogue to Digital Conversion requires a low-pass analogue electronic filter named the "anti-aliasing filter" as described by the sampling theorem.
• The sampled input signal must be bandlimited to prevent aliasing, which means waves of a higher frequency being recorded at a lower frequency
• Likewise, a low-pass filter is required in Digital to Analogue Conversion to prevent aliasing, in this occurrence waves of a lower frequency being recorded at a higher frequency

Implementation

• Sine waveforms have infinite signal responses, both negatively and positively a practical filter is require as it has a non flat frequency
• Some systems have an anti-aliasing filter and a reconstruction filter. They are often identically designed due to input and output both being sampled at the same frequency, 44.1KHz.
• Both attempt to block sounds above 22KHz and as far as possible pass sounds below 20KHZ.
• Theoretically, Digital to Analogue conversion is a series of impulses but is better described as a series of stair steps.
• The low pass reconstruction filter evens out the gap between the metaphorical stairs, remove the harmonics below the required limit

Sound Cards

• A poorer quality of sound card can result limitations in sampling rates, this can be particularly clear in devices such as Notebook PCs
• Most modern sound cards have a 16 bit length, meaning they can represent 2x16 values (65536) different signal levels within the input voltage range
• Quantisation step size can be worked out by dividing the word length by the range.
• Q = 10/65536 = 0.15MV
• The above is a calculation for the range of 10V