A Review of Modern Deepfake Detection Methods: Challenges and Future Directions

Authors:

Zaynab Almutairi

Where published:

Algorithms

Dataset names (used for):

• The M-AILABS Speech: AD detection, Baidu Silicon Valley AI Lab cloned audio: Neural voice cloning with a few samples
• Fake oR Real (FoR): for synthetic speech detection
• AR-DAD: Arabic Diversified Audio
• H-Voice: used to train a machine learning system to classify original and fake voice recordings obtained with the imitation and Deep Voice algorithms
• ASV spoof 2021 Challenge
• FakeAVCeleb
• ADD

Some description of the approach:

The article reviews existing audio deepfake (AD) detection methods and compares faked audio datasets. It introduces types of AD attacks and analyzes detection methods and datasets for imitation and synthetic-based deepfakes.

Some description of the data (number of data points, any other features that describe the data):

The paper focuses on the review of methods rather than specific dataset metrics. The datasets discussed vary in type and size depending on the study.

Keywords:

Audio Deepfakes (ADs); Machine Learning (ML); Deep Learning (DL); imitated audio

Instance Represent:

Various types of audio samples including real and synthetically generated voices.

Dataset Characteristics:

Varied, depending on the study and dataset discussed.

Subject Area:

Audio Security, Machine Learning

Associated Tools:

Detection of audio deepfakes.

Feature Type:

Audio features such as Mel-spectrograms.