Weekly Progress Report
2000-04-06
The short of memory in the DSP seems to be no problem since we already has stored almost as much as we need and we have no problem so far.
After validation of the detection part we have almost implemented the
first complete version of the solution. Some details are left like: matrix
multiplication, toeplitz matrix, the complete AR-library, etc. The big
problem left is that the sound we generate in the DSP doesn’t sound a bit
like the sound generated in Matlab. So far we have no idea what so ever
why.
c) Our first complete version of the solution in Matlab is done and
works correctly. We have also started to modify and improve this version.
We have built a detection method that can detect whether it’s mostly talk
in the sound or not before the beep. If there is mostly talk we should
replaced the beep with a very silent sound.
Not all the fixed AR-parameters are decided optimally. But all documentation
of the solution is done except the detection part.
|
|
|
|
|
|
Find out why the solution in the DSP doesn’t agree with the solution in Matlab. Fix the problem! |
|
|
|
Weekly report written and web page updated. |
|
|
|
Document the detection theory. | |
|
|
Build the last parts to have a first version of the complete solution in the DSP. This mainly means make matrix multiplication in C-language. | |
|
|
Complete the first version of the solution in the DSP, in a way that would be satisfying to present at the end of the course. | |
|
|
Write a first version of the final report. | |
|
|
Improve our complete solution with:
· The replacement sound should have no constant variance. · Adjust our solution so it will work when you don’t know the frequency of the beep. |
|
|
|
Build test cases that prove that our solution is the greatest ever. | |
|
|
Make presentation material:
· Slides in Power Point. · Posters for trade fairs. · Practical presentation (test case). · Webb presentation. |