Detecting Audio Splicing Using MATLAB and Machine Learning
This repositiory contains my files for my auido splicing experimenation. The details for this experiment are given below.
Audio Splicing is a technique of attaching different pieces of audio together to create a new edited audio. Fake audios have become a trend now, and with the increasing expertise of the editors it has become impossible to differentiate between fake and real audio without the use of proper tools.
Also, in the field of Forensics Science, audio forensics is done to validate if a piece of audio is admissible in courts or other official venue. To validate audios different techniques has been used, one of them is using Fourier Transform to detect any splicing. In this project, for detecting splicing in each audio python libraries have been used to analyse the audio, the Machine Learning model have been run to classify different sections of the audio based on their frequencies. Then Fast Fourier Transform was taken, and spectrogram of the audio signal was made to further analyse the audio. And finally, audio toolbox of MATLAB was used to detect splicing in the audio, the number of splices combined, and their duration was recorded. The Machine learning library used with pre trained model was opensoundscape (pytorch models). The pretrained model was RESNET 18.
The project is structured as follows:
Audio-Splicing
├── test/ # Contains test files
├── Python Expriment files/ # Contains Experimentation files
├── Spectrogram Images/ # Contains Images for Spectrograms
├── report/ # report file
├── LICENSE # license file
├── README.md # readme fileTo get started with downloading this repository, follow the steps below:
Clone the repository to your local machine using the following command:
https://github.com/ahmd-mohsin/Audio-Splicing.git
For any further recommnded changes and collaborations, feel free to contact. With ❤️ Ahmed.