With help from the program SumOfSignals, I created an audio signal with three different sinus components. The three sinus components all have unique amplitude and Hertz values.
| x(t)=Σaksin(fk*2πt) |
|---|
| 5000,000*sin(500*2πt) + 3000,000*sin(300*2πt) + 2000,000*sin(200*2πt) |
| Sinus-comp. | Amplitude | Hz |
|---|---|---|
|
|
5000,000 | 500 |
|
|
3000,000 | 300 |
|
|
2000,000 | 200 |
The Signal in D/A screen projects the amplitude over time, wheras the frequence spectrum projects the frequency over time. Looking at the graph, you can easily determine which sinus-component is which by comparing the amplitude to the peaks of the sinus curves. Henceforth, Sinus-component 1, which has the highest amplitude, will have the highest peak in the graph. The coherence between amplitude and the peaks in this graph is obvious as you well can see.
This is a graphical outline of tale1.wav where time and amplitude is represented respectively in the x- and y-axis. Due to the extremely short timeline in the x-axis, it is almost impossible to determine what is being said during that very short period (0,003 sec), but from looking at the graph, I was able to determine a correlation between the pitch level in the voice and the amplitude, i.e. high pitch equals high amplitude, an argument which is also valid inversely.
Whenever you want to play, edit or transfer an audio signal through a computer, you will need to convert the signal from its original analog state to a digital state, a process which we call sampling. This process involves extracting values with certain intervals, and when recreating the signal after the sampling, the computer will fill in the blanks between the derived values, thus creating a new audio signal. If you want a digital audio signal to be accurate to the original, you will need to have proper spacing between the intervals and derive a sufficient number of values. If you fail to provide the computer with enough values, you will end up with a distorted audio signal, which is usually referred to as Aliasing.
The main difference between tale1.wav-tale4.wav is the bandwidth which these signals have been recorded with. The bandwidth determines how much data is provided in a playback of these wave files. If storage space is an issue, you can assign less bandwidth to an audio file, but at the cost of quality.
| file | Listen | Bandwidth | Graph |
|---|---|---|---|
| tale1.wav |
 |
|
|
| tale2.wav |
|
|
|
| tale3.wav |
|
|
|
| tale4.wav |
|
|
|
It is a simple task to determine the diversity of quality between these four audio files. Tale1.wav, which offers the highest quality, provides a smooth and comfortable listening experience with no noise or distortion. Tale2.wav also meets the demand of an unproblematic listening experience; however the difference from its predecessor is noticeable. In tale3.wav you can clearly see the effects of quantification. The effects are also quite apparent in playback, where the sound can be experienced as an unnatural, high-pitched, distorted noise, almost insulting to the ear. And, of course, the same arguments validate tale4.wav's inadequacies, but only to a greater extent. You can clearly hear that the audio signal is computer generated.
Sources: http://www.ifi.uio.no/~inf1040/foiler2003/lyd1.pdf
Sources: http://www.ifi.uio.no/~inf1040/foiler2003/lyd2.pdf
Information Technology : Inside and Outside. Cyganski; Orr; Vaz
