Made a Stone Strider app that basically has Rocky from Project Hail Mary.
You can dress it up in that suit, pick different voices, write whatever you want, and make him say it.
Just a fun thing that came to my mind. Give it a try!
https://t.co/o1DX3a6SkE
I just tried Sarvam's dubbing in their Creator Studio, and honestly this changes a lot for indie devs who want their content to reach people globally and in their native language.
I took one of my explainer videos, made with Gemini Carbon and Droid, and dubbed it into Hindi using @sarvam. It came out flawless and stayed perfectly in sync with the animation.
I think this is the start of my content creation arc. I made a Bengali version too, you'll find it below.
Reading a document is easy. Turning it into data your application can actually use? That's the hard part.
From invoices to forms and government records, extracting structured information from documents can be challenging, especially when dealing with complex layouts, tables, and scanned files.
That's where Sarvam Document AI comes in.
Parse or Extract? Which one should you use?
Parse: When you need the complete document content.
Extract: When you need specific information from it.
Whether you're building AI agents, document search systems, or automated processing pipelines, Sarvam Document AI helps turn documents into usable data.
Explore the documentation: https://t.co/KSzhxZgE9V
Reading a document is easy. Turning it into data your application can actually use? That's the hard part.
From invoices to forms and government records, extracting structured information from documents can be challenging, especially when dealing with complex layouts, tables, and scanned files.
That's where Sarvam Document AI comes in.
How do they work together?
Imagine processing thousands of documents daily.
Split: Separate documents into individual files.
Classify: Identify document types.
Parse: Extract complete document content.
Extract: Retrieve specific fields for databases, applications, or AI workflows.
Both Parse and Extract follow an asynchronous workflow:
Submit → Get job_id → Check status → Retrieve results.