We came together with a single purpose: to apply the mathematics we had been studying in school to benefit humankind. Also, that scene from mission impossible where Ethan Hunt beats Phillip Seymour Hoffman and mimics his voice by recording his voice was cool and we wanted to bring that to reality.

What it does

It differentiates one person's voice from another and can mimic another person's voice through ongoing highly adaptive machine learning algorithms.

How We built it

We utilized Fourier Analysis, Decision Trees, Machine Learning, Signal Processing and Doritos.

Challenges I ran into


Accomplishments that I'm proud of

Complete Interface, converting code and information from MATLAB to C++ to web.

What I learned

Lot of linear algebra, signal processing

What's next for VoiceSingularity

Improvements on the algorithm and mimic feature

Share this project:


William Widjaja posted an update

% Test Project Database(Matlab) %Anurag Shivam Prasad clc; clear all; close all; r1 = audiorecorder(44100, 16,1); disp('Start recording') recordblocking(r1, 5); disp('End of Recording.'); y1 = getaudiodata(r1)'; r2 = audiorecorder ; disp('Start recording') recordblocking(r2, 5); disp('End of Recording.'); y2 = getaudiodata(r2)'; N = length(y1); fn=linspace(-(N-1)/N/2,(N-1)/N/2,N); fn = (fn.*(8000))/(2*pi); y11 = 1/N*fft(y1); y111 = real(fftshift(y11)) y22 = 1/N*fft(y2); y222 = real(fftshift(y22)) v1 = y111/norm(y111) v2 = y222/norm(y222)
v = zeros(1,length(v1)); v = zeros(2,length(fn)); v(1,:) = fn(1,:); v(2,:) = v1(1,:);

a(1,:) = v(1,20000:40000); a(2,:) = v(2,20000:40000);

dt = 1/44000; I0 = round(0.1/dt); Iend = round(0.25/dt); x = y1(I0:Iend); c = cceps(x);

t = 0:dt:length(x)*dt-dt;

plot(t(15:75).*1e3,c(15:75)) xlabel('ms')

[~,I] = max(c(15:55)); fprintf('Complex cepstrum F0 estimate is %3.2f Hz.\n', 1/(t(I+15)))

Log in or sign up for Devpost to join the conversation.