I have an input signal in time domain (that is, discrete 14100 signed integers per second).
If I reverse the polarity (signed-ness) of each value, will I not produce a destructive source-audio canceling signal?
Asked
Active
Viewed 462 times
0
-
Well **1\*X + -1\*X = 0**, so Yes. – RBarryYoung Apr 24 '12 at 02:11
-
1Possible duplicate of http://stackoverflow.com/questions/10290539/polarity-vs-phase-shift-in-noise-reduction – hotpaw2 Apr 24 '12 at 02:46
2 Answers
2
Destructive cancellation by simply reversing polarity will occur only if your total software/hardware/system/transducer delay time from your input signal to to your output signal to the point of cancellation is zero (or a very tiny percentage of the period of the highest frequency), and the frequency and phase response of the destructive transducer is absolutely flat.
e.g. only likely in fictional/theoretical models.
hotpaw2
- 70,107
- 14
- 90
- 153
-
So what do people do in practice? I mean, I do not need the the formula, but rather the concept. What are people accomplishing by manipulating the signals in frequency domain for instance? – user605957 Apr 24 '12 at 04:12
-
Sound moves very slowly in electronic terms. Based on the placement of your microphones and speakers, you may be able to cancel it by *adding* an appropriate delay to your audio. [Autocorrelation](http://en.wikipedia.org/wiki/Autocorrelation) may allow you to determine this delay automatically -- and it is most easily computed by using the frequency domain. – comingstorm Apr 24 '12 at 17:13
1
Yes, Phase inversion (and summing the two) will cancel the signal, resulting in silence. Of course, you must mind numeric limits.
justin
- 104,054
- 14
- 179
- 226