Normalize the amplitude of a sinewave of varying amplitude and frequency

I have a series of sine waves of varying amplitude and frequency, and I need to normalize them to -1 to +1 in amplitude, but maintain the frequency and not induce a phase shift. The idea being that I need to take the arcsine of this signal, and then the slope of that arcsine to get a 0 to 360deg output sawtooth wave. Such that I can compare them to each other to determine static phase offset, and dynamic phase offset (latency). I'm struggling to get a clean normalization, and pretty sure I'm just not fully understanding the "normalize" function. I feel like this might be a fairly normal function, but just don't know where/what to search for. In my mind I think I need to determine each "cycle" by detecting zero crossings and the peaks and valleys, then determining the peak and vally for that cycle, then normalizing to -1 to +1, but repeat it.

Answer 1

[Image Analyst]

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

In the meantime try rescale

normalizedSignal = rescale(signal, -1, 1); % Global rescaling

This assumes the peaks are all about the same height. If you have a "warbling" or "chirp" signal where the peaks of the sine waves vary over time, then look at the envelope function and divide by that envelope.

Answer 2

[Star Strider]

‘I need to determine each "cycle" by detecting zero crossings and the peaks and valleys, then determining the peak and vally for that cycle, then normalizing to -1 to +1, but repeat it.’

If you just want to normalise the amplitudes in every cycle, try this —

Fs = 1000;
L = 10;
t = linspace(0, L*Fs, L*Fs+1)/Fs;
s = sin(2*pi*t*0.75) .* cos(2*pi*t*0.05) * 1.5;

figure
plot(t, s)
grid
xlabel('t')
ylabel('s(t)')
title('Original Signal')

zix = find(diff(sign(s)));                                      % Approximate Zero-Crissing Indices
Ls = numel(s);
for k = 1:numel(zix)-1
    idxrng = max(1,zix(k)-1) : min(Ls,zix(k)+1);
    xv(k) = interp1(s(idxrng),t(idxrng),0);                     % 'Exact' Zero-Crossings
end

dt = t(2)-t(1);                                                 % Sampling Interval

for k = 1:numel(xv)-1
    xrng{k,:} = xv(k) : dt : xv(k+1);
    ys{k,:} = interp1(t, s, xrng{k});                           % Extract Signal Segment By Interpolating Independent Variable
    mx = max(ys{k});
    mn = min(ys{k});
    yv{k,:} = ys{k}/(mx-mn);                                    % Normalise Signal Segment
end

figure
hold on
for k = 1:size(xrng,1)
    plot(xrng{k}, yv{k})
end
hold off
grid
xlabel('t')
ylabel('s(t)')
title('Cycle-Normalised Signal')

This breaks the signal into half-cycles and normalises each half-cycle.

.

.