Detection and identification of a required keyword within an audio content

In modern era of communication, information sharing is very easy and within reach of every common man. Hence, spreading or sharing of ideology is widely possible in very quick time and creates a huge benefit in real time information sharing. With technology there could be a huge possibility of impacting people with harmful information which cannot be tracked. Data privacy is an important factor hence tapping the voice information or monitoring the information becomes illegal so we propose a method based on voice to text conversion and then performing data filtration. The proposed method converts voice to text and looks for illegal words as described by admin and reports the same with number of occurrence of the words with time stamp. The paper proposes a Smart Data Filtration (SDF) technique and extracting Mel frequency and other time domain statistical parameter associated with voice signal. The proposed system was tested on 102 samples of 20 seconds each, where the proposed methodology has shown a high efficiency in tackling the problem associated with violence and hatred speech sharing.