Deep fake audio and video technologies can be used separately and together. There are different use cases for each and their combined uses. However, the combination can create more convincing results in the end products.
Within the computer vision field, deep fake videos use artificial intelligence technologies including machine learning, deep learning, neural networks, generative adversarial networks, and autoencoders.
I identified valuable use cases and valid concerns when conducting design thinking workshops for deep fake technology solutions. The concerns revolve around safety, security, damaging reputation, fake news, mass deception, and blackmails resulting in financial loss.
Non-consensual insertion to videos belonging to politicians, executives, and celebrities is the most common. These public figures can be strategically targeted on social media. And their fake videos might even leak to traditional media if initial diligence is not applied.
While it is challenging to change the actual meaning of a video via basic tools, it is much easier using AI technologies adding emotions mimicking human traits such as face recognition from photos and videos. Several research institutes created remarkable samples using deep learning techniques.
A simple technique is reducing or slightly increasing the speed of speech in a video. However, even this basic technique can confuse watchers giving an altered impression of a person who is not disabled in reality. We have seen some examples of such altered videos on YouTube.
Editing and rearranging text in audio files with speech fonts and markups are straightforward and quick using software packages. Proprietary tools are usually expensive, but some tools in the public domain are freely available to anyone.
Some social media sites have tools to identify fake videos by comparing them with the original ones. But, unfortunately, until they discover them, the damage of misinformation has already occurred, and the compensation usually would not be possible.
There are some tools to change robotic performance to a natural one in video games and movies. Modulation is one of the techniques to achieve this goal. From a technical perspective, the more footage a video has and the larger datasets it uses, the stronger the conceivable outcomes can be.
While initially deep fake applications and tools were produced by propriety technology companies, many open-source communities now produce them. For example, there are more than 80 deep fake open-source projects on Github.
Deep fake technology can be beneficial for various purposes, such as education and training. It can also be ideal for empowering disabled people to share their content effectively. For example, we know that Stephen Hawking shared his valuable messages via computerized tools as he lost his voice due to ALS disease.
Movies and video games can be made more attractive with contributions of deep fake technologies. For example, inserting an unusual character into a film or game which is logistically impossible is a value proposition in the media and entertainment industry.
Business leaders in international organizations can share their content in multiple languages using their faces. So, for example, a CEO of a large corporation can deliver the same speech in various countries in their native langue. Changing the foreign accent in speech is also a valuable use case in multinational teams.
And educators use these types of videos for inspiration and motivation in teaching and learning process.
About the Author
I am a technologist, postdoctoral researcher in cognitive science, author of several books, editor of publications, and digital marketing strategist with four decades of industry experience.
I write articles for Medium, NewsBreak, and Vocal Media. On Medium, I established ILLUMINATION, ILLUMINATION-Curated, ILLUMINATION’ S MIRROR, ILLUMINATION Book Chapters, Technology Hits, SYNERGY, and Readers Hope publications supporting 11,800+ writers on Medium.
You may subscribe to my account to be notified when I post on Medium and join my publications requesting access here. I share my health stories on my publication, Euphoria. My health, mental health, and fitness collection.
Leave a Reply