??? Subscript indices must either be real positive integers or logicals.

Hello. Could someone help me with this please. This code works fine until I change either the frequency values freq1 and freq2 and/or the values in the r=round((freq1+freq2)/(freq2-freq1)+1)/round((freq1+freq2)/(freq2-freq1)-1); to values other than unity i.e. introducing 2* in r=round(2*(freq1+freq2)/(freq2-freq1)+1)/round(2*(freq1+freq2)/(freq2-freq1)-1); sometimes gives the error ??? Subscript indices must either be real positive integers or logicals.

Here is the full script.

fs = 48000;
buf_dur = 1;
freq1=880;
freq2=880*(81/64);
r=round((freq1+freq2)/(freq2-freq1)+1)/round((freq1+freq2)/(freq2-freq1)-1);
[p_1,q_1]=numden(sym(r));
p=double(sym(p_1));
q=double(sym(q_1));
buf=sin(linspace(0,2*pi*freq1*buf_dur,buf_dur*fs))+sin(linspace(0,2*pi*freq2*buf_dur,buf_dur*fs));
plot([0:fs*buf_dur-1],buf)
hold
sound(buf,fs)
n = (freq2-freq1).*freq1.*buf_dur;
k = round((freq1+freq2)*abs(((4*(1:n)+1)-2)/(4*(freq2*q-p*freq1))));
m = round((freq1+freq2)*abs(((4*(1:n)-1)-2)/(4*(freq2*q-p*freq1))));
t_1 = (2*m - 1)/(2*(freq1+freq2));
t_2 = (2*k - 1)/(2*(freq1+freq2));
t_1a = t_1(find(t_1 < buf_dur));
t_1b=round(t_1a*fs);
plot(t_1b,buf(t_1b),'.r')
t_2a = t_2(find(t_2 < buf_dur));
t_2b=round(t_2a*fs);
plot(t_2b,buf(t_2b),'.r')
axis([0 10000 -2.5 2.5])
buf2=zeros(1,fs*buf_dur);
buf2(round(t_1b))=2;
buf2(round(t_2b))=2;
sound(buf2,fs)
sound([buf,buf2],fs)

I've managed to determine that the error starts at t_1a = t_1(find(t_1 < buf_dur)); I'm still learning the basics so any help would be greatly appreciated.

Many thanks

[Walter Roberson]

Which statement does the error occur on?

[Image Analyst]

rick, I see you tried to edit your posting. When you edit, you don't need to have blank lines between lines to get them to show up on separate lines. Just paste in your code, highlight it all, and click the {}Code button above. It will look fine if you do that. I've fixed it for you.

Answer 1

[Image Analyst]

If this is your code

t_1a = t_1(find(t_1 < buf_dur));

the index would be

find(t_1 < buf_dur)

but that returns either integers or empty. but the t1_a line would still execute even if it's empty (I tried it with sample data). So we'd need more info. Are you using the debugger? If you knew how to use that, you'd have this solved very quickly.

Answer 2

[Roger Stafford]

In the definition "t_1a = t_1(find(t_1 < buf_dur));" you have ensured that no index in 't_1b' could exceed 'fs' which is the length of 'buf', so the error cannot be due to an excessive index in "buf(t_1b)". However, I see nothing here that prevents a zero value in 't_1b', which would also produce an error message there. A zero would occur there if an 'm' value were zero, or if an 'm' were 1 and freq1+freq2 were too large (a number less than one-half will round down to zero.)

Roger Stafford

Answer 3

[rick Ballan]

Thanks very much everybody,

Yes you were right Roger and Walter. The problem occurred at

k = round((freq1+freq2)*abs(((4*(1:n)+1)-2)/(4*(freq2*q-p*freq1))));
m = round((freq1+freq2)*abs(((4*(1:n)-1)-2)/(4*(freq2*q-p*freq1))));

which for some frequencies gave k and m as zero for n = 1. A simple change to 4*(2:n) seems to have solved the problem.