06 - Digital Signal Processing Flashcards
What does DSP stand for?
Digital Signal Processing
What are some examples of where you might find DSP being used?
Everywhere! Hearing aids Otoacoustic systems Audiometers Aural rehab software ABR's Cell phones Voice over internet CD/DVD/DAT players MP3 players Biomedical monitoring equipment...
How does DSP compare to Analog (e.g. time, amplitude, etc)?
Analog:
- continuous in time
- continuous in amplitude
- circuits deal with continuous voltages and currents
DSP:
- discrete in time
- discrete in amplitude
- circuits deal with “1’s” and “0’s”
Analog systems transmit signals from the ______ domain, to the ______ domain, back to the ______ domain
Acoustical
Electrical
Acoustical
Name one reason digital systems are so popular?
- Programmability: instructions can be manipulated “on the fly”
- Flexibility: don’t need to rebuild the circuit, like you would with analog
- Advanced signal processing: multichannel compression, precise frequency shaping, feedback cancellation, noise reduction, directional processing
- Features like bluetooth
What domains to digital systems cross when processing an acoustic signal?
Acoustic -> Electrical -> Digital -> Electrical -> Acoustic
How do we discretize time of an analog signal?
Sampling
What does “quantization” refer to?
The discretization of the amplitude of a signal
- the sampled values are converted into bit representation
- the performance of a quantizer is dependent on the number of bits (also called bit resolution)
As we increase the frequency of our signal, we need a _______ (slower/faster) sample rate
Faster
Describe the Nyquist Sampling Theorem
The sampling rate must be more than 2x the highest frequency of the input signal, otherwise there will be distortion
Will a signal processor be able to undo the distortion (aliasing) caused by undersampling?
No
What is “aliasing”?
A waveform that is caused by undersampling, and is not actually part of the input signal
- the digital signal processor cannot tell if these come from the true signal or not
How do we follow the Nyquist criterion if the highest frequency of a signal is unknown?
Anti-aliasing Filter:
- Use a low pass filter to remove unwanted frequencies
- Set the sampling rate greater than 2x the bandwidth of the low pass filter
Are anti-aliasing filters part of the analog or digital circuit?
Analog - they are low pass filters applied before sampling (digital domain)
Why don’t we just increase the sampling rate to get better quality phone calls, hearing aids, etc?
Cell phones and hearing aids are working in real time, so the processor has to deal with that many samples each second
-as we increase the sampling rate, it puts constraints on the processor
Put the following items in order (lowest to highest), based on their typical sampling rate: CD player Cell phone DVD player Hearing Aid
Cell phone (8,000 samples/sec) CD players and Hearing Aids (44,100 samples/sec) DVD players (96,000 samples/sec)
How many “bits” equal a “byte”?
A string of 8 bits = a byte
What does the binary string 1101 equal?
1 x 2^3 = 8
1 x 2^2 = 4
0 x 2^1 = 0
1 x 2^0 = 1
= 8+4+0+1 = 13
Every time we add one more bit, the number of possibilities (“levels”) _______ (stays the same/doubles/triples)
Doubles
E.g.
2^3 = 8
2^4 = 16
8 x 2 = 16 possibilities
Increasing the number of bits increases the ______ of the quantizer
Resolution
In quantization, the # of combinations = 2^b where b = _______
b = the number of bits
e.g. 3 bit = 2^3
When we’re reverse mapping our digitized signal, the step configuration of the signal is essentially made up of high frequency changes (directly up between 2 points), so how to we reconstruct our analogue wave to smooth it out?
Use low pass filter (high frequency steps are filtered out)
Describe the following characteristics of an A/D converter:
Input range
Resolution
Input range - the voltage range that the A/D converter can handle
- can be unipolar (+ve or -ve voltages) or bipolar (+ve and -ve voltages)
Resolution - represented by the number of bits (2^N where N=# of bits)
Match the following quantization values to their corresponding modern device:
Telecommunication system, DVD player, Hearing aid
8 bits/sample, 16 bits/sample, 24 bits/sample
Telecommunication system: usually 8 bits/sample
Hearing aids: 16 bits/sample or better
DVD player: 24 bits/sample
If an analogue to digital converter was operating in real time and was sampling 8000 samples/second with a quantization of 8 bits/sample, how many bits would it be dealing with per second?
8000 samples/sec x 8 bits/sample = 64000 bits/second
True or false: the complete A/D/A setup consists of:
Anti-alias filter -> S/H (sample and hold) -> ADC -> DAC -> S/H -> Reconstruction Filter
True
Name one commonly used signal processing technique
Fast Fourier Transform (FFT) for spectral analysis and transfer function measurements
Digital filtering and filterbanks for frequency shaping, compression, and noise reduction
