Instant voice cloning by MyShell. Contribute to myshell-ai/OpenVoice development by creating an account on GitHub.

Pointer's platform is  dedicated to providing insights and resources for software developers and technology enthusiasts, focusing on programming languages, frameworks, and development tools. Through articles, tutorials, and coding challenges, Pointer offers insights into building software solutions and advancing programming skills. Developers can learn about new programming languages, coding techniques, and best practices in software development.

Pointer

OpenVoice by MyShell offers accurate tone color cloning, flexible voice style control, and zero-shot cross-lingual voice cloning. Join the OpenVoice community for valuable discussions and collaboration opportunities.

myshell-ai/OpenVoice: Instant voice cloning by MyShell.

Interesting concept, but the voice stills sounds unnatural. There were a few sentences where the generated voice put unnatural emphasis on the end of a sentence (e.g. “Ethical concerns surround AI’s societal IMPACT”–you can really hear it from the generated child’s voice).
In most cases, at least to me, it still sounds just as robotic as the AT&amp;T natural voices you hear when you speak with an automated phone system.
Projects like these have several variables to solve for. Emulating emotions is a combination of pitch, tonality, and tempo. But not all emotive text should be interpreted with the same combinations. For example, their “Sad” example sounds like a voice actor who’s phoning it in and not someone who is actually sad.
I can see the technology improving as the project continues to evolve. I question how ethical it is to sample someone’s voice and generate speech on their behalf.