bob.paper.lipsync2019 1.0.0

Source code for EUSIPCO 2018 paper on audio-visual inconsistency detection

Speaker Inconsistency Detection in Tampered Video

This package is part of the Bob toolkit and it allows to reproduce the experimental results published in the following paper:

@inproceedings{KorshunovICML2019,
     author = {Korshunov, Pavel and Halstead, Michael and Castan, Diego and Graciarena, Martin and McLaren, Mitchell and Burns, Brian and Lawson, Aaron and Marcel, S{\'{e}}bastien},
   keywords = {inconsistencies detection, lip-syncing, Video tampering},
      month = jul,
      title = {Tampered Speaker Inconsistency Detection with Phonetically Aware Audio-visual Features},
  booktitle = {International Conference on Machine Learning},
     series = {Synthetic Realities: Deep Learning for Detecting AudioVisual Fakes},
       year = {2019},
}

If you use this package and/or its results, please cite the paper.

Installation

The installation instructions are based on conda and works on Linux and Mac OS systems only. Install conda before continuing.

Once you have installed conda, download the source code of this paper and unpack it or checkout from Gitlab. Then, you can create a conda environment with the following command:

$ cd bob.paper.lipsync2019
$ conda env create -f environment.yml
$ source activate bob.paper.lipsync2019  # activate the environment
$ python -c "import bob.io.base"  # test the installation
$ buildout

This will install all the required software to reproduce this paper.

Documentation

• Download databases and generate tampered videos

• Reproduce the experiments

Contact

For questions or reporting issues to this software package, contact Pavel Korshunov ([e-mail]).