VoxForge
VoxForge
Hi everyone,
I'm asked to write a program functioning like this:
Given some sound recordings(recorded at street, might have noise or other unrelevant sound), some people have labelled out the CAR ENGINE SOUND in timestamps. I want to use these labelled recordings as training set to train a model, then given a new recording, I can use the model to auto label any CAR ENGINE SOUND timestamps in it.
sounds like a pretty easy job, but I really don't have any experience on handling time series signal. Can someone give me some ideas ? I'd appreciate it very much!
--- (Edited on 3/24/2016 8:58 am [GMT-0500] by taa199) ---
You might want to check out the field of machine learning. When you are given a labelled data set as a training set there are various tools available for building a model which you can then apply to a test set to make a prediction about the nature of the test set.
Coursera and other institutions offer basic free courses in machine learning. By following such a course you would get an appreciation of what tools might be appropriate in your specific case.
--- (Edited on 2016-03-29 6:14 am [GMT-0400] by colbec) ---