Independent Component Analysis is a technique that uses the principles of the Blind Source Separation technique. ICA has a wide range of applications in many fields like medicine, films, analytics, sound recognition etc.
Blind Source Separation is a classical problem when a signal is mixed with a variety of undesired noise and the source is unknown, in a case like this our goal is to understand the source of all the individual components and separate the source.
To understand the cocktail party problem one has to imagine themselves at a party with multiple people, naturally one can imagine a variety and mixture of sounds from unknown sources this is often referred to as “complex auditory setting”. We have to imagine shutting all other voices out and listening clearly to sound generated from a single source. This type of problem can be solved by applying the well-known algorithm- ‘Independent Component Analysis’.
Here I have implemented the fastICA using sklearn To Use- Change path variables and add desired sound file in .wav format