AI generated voice resembles the real voice of Joe Rogan

Joe Rogan said in his latest video that he wanted to sponsor a hockey team made up entirely of chimpanzees. Moreover, he claimed that he is under the robot's control. However, it was not the voice of Rogan, it was AI generated voice which can resembles the voice of any comedian or podcaster.
This AI-Generated Joe Rogan Voice Sounds So Real It's Scary


The dupe of The Joe Rogan Experience’s host voice was created by artificial intelligence company Dessa and its machine learning engineers who used the host's voice to make tools to learn the voice patterns and style, according to the video description. This AI can help the creators, podcasters and YouTubers who produce a ton of content for their audiences on a frequent basis.

The company, on the other hand, explained that AI was able to generate the resembling voice because engineers wrote sophisticated codes — "Using a text-to-speech deep learning system they developed called RealTalk, which generates life-like speech using only text inputs" — due to which the machine is able to produce replica of anyone's voice by using the data stored in it.

Furthermore, Dessa's spokesperson has also said that AI was trained on the audio of Rogan and what was listened by viewers was the result in of text inputs.

This program is like that algorithm which made Jordan Peterson's rap to sound like Eminem. However, that algorithm needed audio of six hours to get trained. Yet, programs, like Modulate.ai and Lyrebird also need some minutes long audio to generate replica.

Thus, Dessa's replica stand out of all, but the company disagreed to release the information of algorithm’s working and codes. Yet, the team promised to post some technical overview about it, in next few days.


Currently, there are many startups which try to make their fake image, after the rise of deepfakes - algorithm based face swap videos, that have some negative impact on society. Dessa highlighted those impact like harassment, impersonation, scammers and scam call, but on the other hand review gave points on how to improve quality of fake audio.

If you think that you can understand the difference between AI generated voice and real voice then play the quiz which scientist has created to know that how much you can differentiate between Rogan's real voice and fake voice. The two voice have slight difference of inflection.

Featured photo: Paul Archuleta / Gettyimages

Read next: Translatotron by Google is Capable of Translating in Speaker’s Voice

2 comments: