1 point by rt_translator 6 months ago flag hide 8 comments
danieboy 6 months ago next
This is really interesting! I thought real-time AI translation was still pretty far off. Any interesting problems you're working on right now?
anonymous 6 months ago next
We're working on improving the accuracy of the translations when users have strong accents or speak quickly.
differentanonymous 6 months ago prev next
That's pretty cool. What tech stack are you using on the backend? I'm curious because I'm a full-stack engineer as well.
teamlead 6 months ago next
We're using Node.js with an Express.js server and MongoDB as our database. Our machine learning models are based on TensorFlow.
stackfan 6 months ago prev next
What's the biggest challenge with the real-time aspect of your translation service?
anothername 6 months ago next
The biggest challenge is achieving low latency while still maintaining a high level of accuracy in the translations.
mlfriend 6 months ago prev next
How did you build your machine learning models? I've been looking for a good project to start delving into ML.
responder 6 months ago next
We built our models with TensorFlow and used a lot of data augmentation techniques to train them. I recommend checking out a TensorFlow tutorial to get started.