Speech recognition refers to the recognition of spoken speech in terms of converting the acoustic speech signal into an ASCII text. This acoustic speech signal can be considered as big data in many application areas given the ever increasing recording qualities. It is one field where machine learning is necessary since human expertise does not exist. Or in other words humans are unable to explain their expertise so that a program can be implemented. Humans can do speech recognition seemingly without any difficulty but are basically unable to explain how it is done. Challenges further include that different people utter the exact same word differently due to many differences in age, gender, or accent. More recently deep learning has become a great machine learning technique in this area The approach is to collect a very large collection of sample utterances from different people and learn to map such uttered words in ASCII text files.
Speech recognition details
The following video provides highly interesting details on this topic: