|
| WaveWarp 2.0 Example DrawingBoard
AdaptiveHumCancellation
Description
Adaptive Hum Cancellation
Illustrates the use of adaptive filtering for narrow-band noise rejection (i.e. hum cancellation). In this
example, a signal generator is used to artificially add excessive low-frequency "power hum" to the audio signal.
The Tracking Peak Detector determines the frequency of the hum based on the combination of STFT (Short Term Fourier
Transform) and phase-tracking techniques (see "TrackingSpectralPeakDetectionEducationalExample.dwb" for more
information). The first control signal output from the Tracking Peak Detector contains the measured hum frequency in Hz
(the second contains its rms amplitude but is unused in this example). This frequency measurement is fed directly into the
"center frequency" control input of the Controllable 2-pole allpass filter. Also, a "bandwidth" control signal is derived from
the "center frequency" by multiplication via the Large Control Gain, and fed to the second control input of the Controllable
2-pole allpass filter. The noisy audio signal is fed into the audio input of the Controllable 2-pole allpass filter. The ouput of
the Controllable 2-pole allpass filter is combined in parallel with the noisy signal, before being sent to the soundcard.
Recalling that when an allpass filter is connected in parallel with the original signal, a notch filter is obtained, it becomes
apparent how this system works: in summary, the controllable allpass filter connected in parallel with the noisy signal has
the effect of "notching out" the noise. The system is adaptive in the sense that the "center frequency" and "bandwidth" of
the notch automatically track the frequency of the noise, by virtue of the Tracking Peak detector.
Experiment with the settings of all components to explore the interesting behaviour of the adaptive filter. For example, try
changing the frequency of the hum, you will notice how the system rapidly "locks on" to the new frequency and cancels
the noise.
It is instructive to observe that when the noise and the signal share the same frequency band, it is more difficult to filter out
the noise without adversely affecting the signal. This is an important lesson in the art of filtering.
See also the "HumCancellationByNotchFiltering" example DrawingBoard for simple (non-tuning) hum removal by manual
placement of the notch..
Components used:
|
|
