AS technology evolves globally, fact-checkers and journalists are confronted with the rising challenge posed by tools which purveyors of disinformation use in creating fake videos, images, and audio that depict individuals saying or doing things they never said or did.
These tools are being used to spread disinformation and commit fraud through the impersonation of public figures or individuals in positions of trust. They have become a powerful tool for malicious actors, making it easier to deceive large audiences and harder for victims to discern the truth convincingly.
Although until recently, the most widespread audio-visual disinformation content are in video format, audio deepfake has become a significant disinformation threat, especially during elections.
READ: How to protect yourself from online scams – Study
For instance, during Nigeria’s 2023 general election, there were audio deepfakes spread to sway public opinion. For instance, a few hours before voting started during the presidential election, a manipulated audio of Nigeria’s former vice president and PDP candidate in the poll, Atiku Abubakar, purportedly discussing plans to rig the election surfaced online.
Similarly, the Labour Party’s presidential candidate in the election, Peter Obi, was in the eye of the storm, after the February 25 presidential election. An audio recording of a telephone conversation that was purportedly made between Obi and the presiding Bishop of Living Faith Church, Bishop David Oyedepo in which Obi was heard canvassing for votes and allegedly used the phrase ‘Religious War. The “yes daddy” audio leak is a 4 minutes, 17 seconds audio clip which had gone viral. But Obi denied the recording, describing it as fake.
In this tutorial, our fact-checker look at some tools that can help anyone detect audio that are synthetic or created using artificial intelligence.
1. Reality Defender
This AI app detects audio deepfake by identifying artefacts and anomalies that arise while synthesizing any audio. The web application allows users to submit audio files for analysis, including videos and images too.
WATCH THIS: Gaining Trust for the Media in the Fake News Era
The app analyzes the file and informs the user when the process is finished via the dashboard and an email alert. Reality Defender’s deepfake detection platform is primarily used by enterprises, platforms, and governments. However, they also offer a way to detect audio deepfakes.
2. PlayHT Voice Classifier
This AI voice detector stands out as one of the top choices for detecting AI-generated voices. This advanced tool leverages cutting-edge machine learning algorithms to accurately differentiate between real and synthetic voices.
PlayHT offers an easy-to-use interface, making it accessible for both professionals and casual users. The tool analyzes various audio features and provides detailed reports on the likelihood of the voice being synthetic or real. It also offers its audio detection service freely for users.
3. Resemble AI Detector
This tool offers a robust voice detection tool designed to identify synthetic voices. It enables robust deepfake detection by combining efficient architecture with unparalleled accuracy across diverse languages and generation methods.
ALSO READ: Cybersecurity firm cautions organisations, individuals against deepfakes increase
Resemble AI Detector is particularly useful for businesses looking to protect their brand and prevent voice cloning scams. Its intuitive interface and detailed analysis reports make it a go-to solution for voice verification needs.
4. Loccus AI
Loccus.ai is a startup building deep-learning solutions to add layers of trust and safety to voice communication. Their product aims to solve the misuse of AI-generated voices aimed at committing fraud, spam and misinformation.
The tool is now added to Hiya’s Adaptive AI fraud prevention system with AI voice detection to provide businesses and carriers with a complete fraud call protection suite. The versatile solution supports multiple languages, formats and platforms, including video recordings and live calls across all devices.
5. Deepware Scanner
Deepware Scanner is an open-source tool designed for detecting deepfake audio-visual content. It uses a combination of machine learning algorithms and spectral analysis to identify synthetic voices and can be used for various applications, from social media content verification to security and forensic analysis. Though it works mostly for videos, it can be used to detect fake audio too.
Its open-source nature allows for customization and continuous improvement by its users’ community.
At this juncture, it is important to note that these tools are not infallible. While they can estimate the likelihood of a file being AI-generated or synthetically created, they are not correct always. That is why they need to be used along with other methods of verification.
Nurudeen Akewushola is a fact-checker with FactCheckHub. He has authored several fact checks which have contributed to the fight against information disorder. You can reach him via [email protected] and @NurudeenAkewus1 via Twitter.
