In Brief
Researchers from Carnegie Mellon University have created a new artificial intelligence model that is able to generate realistic speech by learning from YouTube videos and podcasts.
The model was able to learn the nuances of natural speech patterns by listening to almost 900 hours of YouTube and podcast content, resulting in a more realistic and lifelike synthetic voice.
The Trust Project is a worldwide group of news organizations working to establish transparency standards.
Researchers from Carnegie Mellon University in the USA have created a new artificial intelligence model that is able to generate realistic speech by learning from YouTube videos and podcasts. The model is capable of producing speech in a variety of voices and accents, making it a potentially useful tool for industries such as entertainment and advertising. It could also have implications for those who rely on assistive technology to communicate.
This is a significant development, as the majority of AI-generated speech has been found to be emotionless and monotonous due to the fact that it is usually trained on datasets where the speech of professional actors is recorded. The new technology can be used to improve the performance of virtual assistants and chatbots, making them more engaging and human-like. It also has potential applications in fields such as gaming, education, and entertainment.
The new model, however, was able to learn the nuances of natural speech patterns, such as pauses, interjections, and parasitic words, by listening to almost 900 hours of YouTube and podcast content. This resulted in a more realistic and lifelike synthetic voice, which was rated 3.89 on a five-point scale by people who were asked to evaluate it. This is higher than the average score received by similar AI models (a real human voice scored 4.01).
The potential applications of this technology are vast and include helping people with speech impairments, improving navigation systems, and creating more natural-sounding virtual assistants.
- PodcastAI is the first podcast ever produced by AI, featuring a spoof of Joe Rogan interviewing Steve Jobs. This podcast is made entirely by AI computer algorithms. The AI was trained for the Steve Jobs episode using both his biography and any internet recording of him that could discover. This allowed the AI to faithfully recreate his personality.
Read more related articles:
Disclaimer
Any data, text, or other content on this page is provided as general market information and not as investment advice. Past performance is not necessarily an indicator of future results.
