<< Chapter < Page Chapter >> Page >

Procedure for echo removal by deconvolution

1. Generate a Gaussian white noise input signal in matlab like the one below:

A sample of the signal can be heard in this file:

2. Record a test speech signal in a low-echo environment like a music practice room. The time domain version ofa speech signal where the speaker is counting to ten can be seen below:

The speech recording can be heard in this file:

3. Record ambient noise characteristics by having a microphone run in the room you want to use while nothingelse is going on.

4. Play the Gaussian white noise signal in the room and record the room’s response.

5. Play the test signal from step 2 and record the room’s response.

6. Deconvolve the room’s response from the Gaussian white noise recording in step 4 by using normaldeconvolution.

7. Deconvolve the test signal from the test recording in step 5 using either normal deconvolution, or Wienerdeconvolution with calculations of the noise characteristics from the noise recorded in step 3.

Get Jobilize Job Search Mobile App in your pocket Now!

Get it on Google Play Download on the App Store Now




Source:  OpenStax, Elec 301 projects fall 2006. OpenStax CNX. Sep 27, 2007 Download for free at http://cnx.org/content/col10462/1.2
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'Elec 301 projects fall 2006' conversation and receive update notifications?

Ask