Show HN: Voice cloning App with 45 celebs' voices made by Tacotron2 and WaveGlow

parodist.ai

5 points·vladimirsvsv77··3 comments

Hey Hacker News!

I’m Vladimir, founder and CEO of the Parodist app. I’m very excited to share our app with you.

Parodist is a text-to-speech mobile app that uses AI technology to sound personalized messages in voices of famous politicians, celebs, and tech leaders.

Our early users love the opportunity to congratulate their friends with the personalized audio of Taylor Swift or prank their family members with the video of Barack Obama endorsing them for the POTUS position. Now, we have 45 voices, and we’re constantly adding new ones.

It’s fun, safe, and easy to use. Just enter any name into the template and get the video or audio voiced.

We’d be glad to hear your feedback and discuss any questions here. Also, you may reach out to us at hello@parodist.ai.

P.S. We’re also open to any kind of b2b inquires as we have simple and fast API. We already have some business cases, such as sounding articles on the website of the popular radio station; sounding messages that come along with donations during streaming; recreation of the voice of the passed away person for the documentary, etc. For more info you may contact us at the email below.

This is very cool, just tried the App :) Homer and Cartman work somehow better than Rick. How does the backend work? And how did you develop the app? Very good job! Congrats for the launch.

Thank you! We hired parodists to collect training data for voices with full access, one parodist did Homer and Cartman, and another did Rick, this is why there is a difference. On the backend, we use a microservices with models on a GPU server (AWS) and a simple web service with Django and Amazon Aurora (serverless). The application was made with react-native for both platforms at once.

Reply on news.ycombinator.com