I read the stack exchange link posted and the answer is quite old. Maybe a more efficient solution can be reached using neural networks for your purposes. However, one question is whether you have any annotated data at your posession for training.
Through the coursework of my master last year, I believe I have the knowhow be able to tackle your problem. However, I cannot give any estimates of accuracy for sure, since I haven’t conducted a similar task..
This is why the first milestone is a detailed proposal where we can reach an agreement about the technical details of what should be implemented and a work plan. This work plan can for sure be executed by someone other than myself. But honestly, I don’t think I can give a concrete quote and time about when this would be implemented. I am aware of what should be done but only have experience at the scale of master coursework projects so I would still have to read and write a technical plan, to be executed by me or another developer (you can decide later for sure)
I am meticulous, and a good communicator. Currently I am doing my masters in Sound and Music Computing in Universitat Pompeu Fabra. Next year is my second. Before, I’ve had 4 years professional experience as a Software Developer, 1.5 of them in C++. I am excited to build experience in tasks related to the content of my masters program which is what I hope to be doing for a while.
So, I hope you reach out to me whether for this or future collaborations :)