EzDubs: Real-time speech translation

Real-time speech translation

EzDubs is a real-time speech translator. Our proprietary models translate speech without waiting for complete sentences, clone voices on the fly (without enrollments), and even replicate emotions in the translated language. Our iOS and Android apps translate live phone calls, automatically dub voice messages, and auto-translate text conversations so that you can communicate seamlessly across languages, with your voice and emotions. Thousands of users use EzDubs to communicate with friends and family who speak different languages, manage offshore teams, and even have relationships across languages. Visit https://ezdubs.ai to download our app and try a live demo of our real-time speech translator, or reach out to explore how EzDubs can integrate with your communication stack to enable live multilingual conversations.

Active Founders

Padmanabhan (Paddy) Krishnamurthy

Founder

Building EzDubs: real-time speech translation. Columbia MS CS (thesis on cross-lingual speech synthesis), HKUST BS in CS + AI (Distinguished Student Researcher, President's Cup Winner, Best Final Year Thesis), ex-Visiting Researcher at USC Interaction Lab, ex-ML at Webex (zero-shot voice cloning), GGCS 2019 runners-up (ML for autonomous ocean cleaning robots) Previously built Helen, an award winning wearable camera that lipreads using AI (James Dyson Award 2019, IET PATW 2019).

Padmanabhan (Paddy) Krishnamurthy

Founder

Building EzDubs: real-time speech translation. Columbia MS CS (thesis on cross-lingual speech synthesis), HKUST BS in CS + AI (Distinguished Student Researcher, President's Cup Winner, Best Final Year Thesis), ex-Visiting Researcher at USC Interaction Lab, ex-ML at Webex (zero-shot voice cloning), GGCS 2019 runners-up (ML for autonomous ocean cleaning robots) Previously built Helen, an award winning wearable camera that lipreads using AI (James Dyson Award 2019, IET PATW 2019).

Amrutavarsh Kinagi

Founder

ML researcher and co-founder of EzDubs, building real-time speech translation for live, expressive conversations. Holds an MSCS degree from Columbia (Speech AI researcher) and a BS from HKUST in CS + Math. Previously co-founded Helen AI, an award-winning lip-reading tool for the hearing impaired (HKUST President’s Cup, IET YPEC, Runner-up Dyson Award Hong Kong). Led HKUST robotics work on autonomous systems (Runner-up, ROBO-ONE Japan; 1st Class Award, NXP Cup 2018).

Amrutavarsh Kinagi

Founder

ML researcher and co-founder of EzDubs, building real-time speech translation for live, expressive conversations. Holds an MSCS degree from Columbia (Speech AI researcher) and a BS from HKUST in CS + Math. Previously co-founded Helen AI, an award-winning lip-reading tool for the hearing impaired (HKUST President’s Cup, IET YPEC, Runner-up Dyson Award Hong Kong). Led HKUST robotics work on autonomous systems (Runner-up, ROBO-ONE Japan; 1st Class Award, NXP Cup 2018).

EzDubs is an AI video dubbing tool that translates videos and livestreams while preserving the voice of the original speaker.

Website (with demo videos)

Mission

Breaking down language barriers to videos across the internet

Language inaccessibility sucks

Enterprises and creators can’t broadcast their videos and livestreams to different geographies, and lose out on brand growth and revenue
Consumers can’t discover foreign language content and recommendation engines keep them isolated within their geography

Current dubbing solutions are:

Expensive - inaccessible to smaller creators / enterprises
Human dependent - having humans translate and voice every video leads to turn around times of days/weeks and isn't scalable
Unable to service livestreams - live interpreter services are expensive, require humans to join each call, and lose the original speaker’s identity

The EzDubs Solution - Fully automated, real-time AI dubbing

Instant turnarounds - dub livestreams in real-time and simultaneously engage with multiple audiences across the globe
Scale + affordability - automatically dub all videos uploaded to a platform using our API and eliminate hiring human interpreters / voice actors for each different language
Voice preservation - preserving the voice of the original speaker in the translated language makes for a more authentic viewing experience (esp. for livestreams), and enables creators to grow their brand internationally

Can I try it?

Yes!!

Twitter Bot: Tag us in a reply to a video and receive the dubbed video in ~3-5 mins
Demo
Chrome Extension: Embeds a “Translate” button below videos on YouTube and Twitter
iOS Shortcut: Call EzDubs from within the YouTube app on iOS

Works best on:

Short-form videos (< 5 minutes)
Videos without much background noise (podcasts, interviews, etc.)

How can you help us?

Do shout out our bot on Twitter, so that we can be discovered by communities (like football / crypto sub-Twitter)
Would be great if you could connect us to publishers / live streamers / content creators looking to grow their international following
If you’re an enterprise that markets internationally, we’d love to talk outreach strategies
Share this post! Let the world know that real-time dubbing is now a reality and no longer Star Trek / Babelfish sci-fi.

Cheers, and happy dubbing!

Paddy and Amrut

YC Photos