This is a submission for the AssemblyAI Challenge : Sophisticated Speech-to-Text.

What I Built

Spext is a web app that helps international students study better when they are struggling to keep up with lectures because of the language barrier. Spext can

• Convert audio to text
• Translate audio to your language
• Convert and translate stream audio to text (Coming soon)
• Sumarize the audio (Coming soon)
• Ask questions about your audio data (Coming soon)

Demo

https://spext-ruby.vercel.app/

Journey

My technical stack is:

• NextJS for UI
• Firebase for authentication
• AssemblyAI’s Speech-to-Text Model APIs to convert audio to text
• An API to translate the language
• Vercel for hosting

I use Universal-2, AssemblyAI’s Speech-to-Text Model APIs to convert an audio to text
Then I create a feature which translates the converted text into the language that the user is fluent. Currently, I support 26 languages

In the future:

• I plan to develop a feature for real-time translation, allowing students to understand lectures as the instructor speaks instead of having to rely on recordings.
• I will support translation into more languages and make it more natural and human-like.
• I’ll create features to summarize audio, ask questions about its content, and generate flashcards based on the audio. <!-- Tell us how Universal-2, AssemblyAI’s Speech-to-Text Model was incorporated in your application. -->