Speech signal power decrease speed

Hi,
I have designed a system/algorithm which attenuates speech signal. Now I would like to calculate using Matlab - from my system's output speech signal (a wav file) - the power of original signal and power of processed signal and see how fast it is attenuated let say -10dB. I expect that would happen in some tens (or hundreds) of msec's. What would be convenient way to calculate the powers and get the result? Best would be if I could plot the powers graphically. Any .m code example for this available?
If I calculate signal powers frame-by-frame (let say 10ms frames), the speech signal is changing too fast and not that visual?
Thanks,
Micke

6 Kommentare

Mathieu NOE
Mathieu NOE am 27 Apr. 2021
hi
why not plot the envelop of the signal ? you should be able to see that drop in amplitude
see : envelope Envelope detector.
[YUPPER,YLOWER] = envelope(X) returns the upper and lower envelopes of
the input sequence, X, using the magnitude of its analytic signal.
Mikael
Mikael am 27 Apr. 2021
That might be a good solution. Just wondering can I plot it so that it will "alert" if output is dropped -10dB or more compared to input...
Mathieu NOE
Mathieu NOE am 27 Apr. 2021
do you have already started to code ? maybe I can jump in if you have a code to share
Mikael
Mikael am 27 Apr. 2021
Great, thanks. I already did a very basic start plotting upper envelopes:
[signalIn, fs] = audioread('input.wav');
[signalOut, fs] = audioread('output.wav');
figure
[YUPPERin, YLOWERin] = envelope(signalIn);
[YUPPERout, YLOWERout] = envelope(signalOut);
plot(YUPPERin, 'r');
hold on;
plot(YUPPERout, 'g');
Mathieu NOE
Mathieu NOE am 28 Apr. 2021
hello
can you share the wav files as well ?
Mikael
Mikael am 3 Mai 2021
Sorry for delayed response but I was able to confirm from time domain signals that my algorithm is fast enough and gave up from this deeper checking as it is challenging to calculate it after all. Thanks for the help.

Melden Sie sich an, um zu kommentieren.

Antworten (0)

Produkte

Tags

Gefragt:

am 27 Apr. 2021

Kommentiert:

am 3 Mai 2021

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by