Echo generator problem - Coursera Introduction to Matlab Programming

Hey all,

I am trying to program the solution to the Echo Blur problem (I have attached a screenshot below describing the task):

My code is written below, could anyone give me any pointers as to where I am going wrong?

function[output] = echo_gen(input, fs, delay, amp)
input = input'; %storing the column vector input as a row vector 
samplesinthesrecording = length(input); %finding the number of samples in the input vector 
numberofdelaysamples = round(delay)*fs; %finding the number of samples in the delay 
sampleatwhichechostarts = numberofdelaysamples + 1; %finding the sample number at which the echo begins
soundbeforecho = input(input < input(sampleatwhichechostarts)); % the sound before the echo is playing during the samples which appear before the echo  
amplifiedecho = input*amp; %the echo is amplified by an amount specified by amp 
originalsoundplusecho = [soundbeforecho, amplifiedecho]; %I am binding the vector of samples for both the original sound and the echo 
if min(originalsoundplusecho) < -1 || max(originalsoundplusecho) > 1 %I am making sure the values within the vector 'originalsoundplusecho' are less than 1 and greater than -1 
   normalisedvector = rescale(originalsoundplusecho,-1,1); %I am normalising the vector if the condition above is not met, making sure it is in the boundaries -1 to 1
   output = normalisedvector'; %the output is the normalised vector 
else 
   output = originalsoundplusecho'; %if the values of the vector 'originalsoundplusecho' are within -1 to 1, then the output is the vector 'originalsoundplusecho'
end 
end 

Here is the code to call the function:

load splat
output = echo_gen(y, Fs, 0.25, 0.6); %calling the function 
dt = 1/Fs; %the interval between values on the x-axis 
t = 0:dt:dt*(length(output)-1);%the x-axis scale 
plot(t, output) %plotting the graph 

Thank you very much!

[Walter Roberson]

What leads you to think that you have errors in the code?

[ey21]

@Walter Roberson, there are no such errors in the code itself, but the program is telling me that my code is not accepted for the following reasons:

Tested with the vector [-0.5; 0; 0.5; 0] and the following parameters: fs = 1, delay: 0.0 seconds, amp = 0.5 - this gave the incorrect output

Variable output must be of size [10820 1]. It is currently of size [16055 1]. Check where the variable is assigned a value. Tested with the splat file and the following parameters: fs = 8192, delay: 0.1 seconds, amp = 0.0 - it also gave me this message.

This is the output the code is giving me: (Screenshot)

[ey21]

@Walter Roberson could you give me any advice on what might be going wrong?

[Walter Roberson]

soundbeforecho = input(input < input(sampleatwhichechostarts)); % the sound before the echo is playing during the samples which appear before the echo

input(sampleetc) is the value at that location. input < that is a logical vector with true for all locations whose value is less than that particular value, including samples before and after the start of the echo. input() indexed at that selects the content of all samples whose value happens to be less than the value at the start of the echo.

I do not understand what possible use such a vector would be for this project.

if you want to select the first 10 samples then input(1:10)

please do not use input as the name of a variable, as it confuses readers who expect input to refer to the matlab function to ask the user to enter data.

[Emre Yavuz]

@Walter Roberson, in the function:

soundbeforecho = input(input < input(sampleatwhichechostarts));

In am selecting the samples which come before the echo from the 'input' column vector.

I am then binding this sample to a vector with the samples within the echo:

originalsoundplusecho = [soundbeforecho, amplifiedecho];

So effectively, the output of the function should be the values of the y-axis which correspond to the sound being replayed and then the echo of the sound (with an amplification).

[Walter Roberson]

In am selecting the samples which come before the echo from the 'input' column vector.

No, you are not. Suppose for example

input = [3 1 4 1 5 9 2 6 5 3 5];
sampleatwhichechostarts = 9;
input(sampleatwhichechostarts)
input < input(sampleatwhichechostarts)
input(input < input(sampleatwhichechostarts))

ans =
     5
ans =
  1×11 logical array
   1   1   1   1   0   0   1   0   0   1   0
ans =
     3     1     4     1     2     3

When what you want is [3 1 4 1 5 9 2 6] which is input(1:8)

[ey21]

@Walter Roberson, thank you very much!

The code below works!

function[output] = echo_gen(input, fs, delay, amp)
numberofdelaysamples = round(delay*fs); 
soundvectorplusecho = zeros(numberofdelaysamples + length(input),1); 
soundvectorbeforecho = soundvectorplusecho; 
for i = 1:length(input) 
    soundvectorplusecho(i + numberofdelaysamples) = input(i)*amp;
    soundvectorbeforecho(i) = input(i); 
end
fullsoundvector = soundvectorbeforecho + soundvectorplusecho; 
range = abs(fullsoundvector); 
maxrange = max(range);
if maxrange>1 
   fullsoundvector = fullsoundvector/maxrange; 
end
output = fullsoundvector; 
end 

[Onkar Arora]

i can't understand

Answer 1

[Muhammad Qaisar Ali]

function output = echo_gen(input, fs, delay, amp)
    delay_mat=zeros(round(delay*fs),1);    % a col vector.
    echo_mat=([delay_mat;input])*amp;          % make echo col vector,input is a column vector.
    output=input+echo_mat(1:length(input),1);       % super imposing echo with origional input sound track.
    output=[output;echo_mat(length(input)+1:end,1)];
    if max(abs(output))>1    % scaling b/w -1,+1 throught relative scaling.   
        output=output/max(abs(output));
    end
end