Speech Forgery Detectionof Framed Sentences In Audio Recordings Using DTW
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 8, Pages 2269-2274
AbstractOur main objective is to detect the speech forgery that is formed by framing sentences. These sentences can act as a crucial evidence in courts and in audio forensics and in NLP. In this these sentences are formed with the help of Telugu alphabets which are audio recorded. The alphabets act as a source and basics to form these sentences. The procedure of forming these sentences is as follows from alphabets to words and words to sentences by concatenation. After forming these sentences with help of MFCC which is feature extraction technique used in audio and speech data extraction. After the extraction we are using dynamic time warping technique to estimate the forgery in the audio which is formed from framing sentences. This forgery detection of framed sentences can be further be used in many ways such as in NLP and speech to text conversion applications in audio forensics and some of the new generation applications. The proposed method is able to differentiate the forgery by comparison from original and forged frame sentences. This is done and implemented with help of MatLab tool.
- Article View: 45
- PDF Download: 114