Vocal script ai

Turn your voice or thoughts into professional scripts and voiceovers.

Comment

Inspiration

Every time I've had an idea for video or content, I've struggled with how to start or what to say. I soon realized many of my friends faced the exact same challenge: they could use AI to generate images, voices, or even videos, but they were stuck on the fundamental – what to actually write or say. This common struggle is what truly inspired me to create Vocal Script AI, a tool designed to effortlessly transform raw ideas into structured scripts and professional voiceovers, making content creation genuinely easier for everyone.

What it does

Vocal Script AI is a tool that transforms a user’s typed or spoken input into a well-structured, ready-to-use script. It also offers an optional voiceover feature — so users can listen to the script, which is especially helpful for those who prefer audio over reading.

How we built it,

Frontend

-HTML, CSS, Vanilla JS: For a responsive and intuitive user interface.

-Rendering Library: To efficiently display dynamic content.

Backend

-Node.js, Express.js: A powerful and flexible runtime environment and web application framework.

-TypeScript: For type-safe and maintainable backend development.

AI Integration:

-HuggingFace: Utilized for advanced text generation and text-to-speech functionalities.

-OpenAI: Integrated to support and enhance specific AI models via HuggingFace.

Key Libraries

-dotenv: Securely manages API keys and configuration settings.

-kokoro-js: Powers our text-to-speech engine.

-onnxruntime-node: A crucial dependency for kokoro-js.

-nodemon, tsx: Dev-time tools for running backend code (e.g., via npm run dev)

Challenges we ran into

-It was my first time integrating AI models in the backend. -First experience using MediaRecorder API.

-Learning to work inside the Adobe Express Code Playground environment.

-Finding free, reliable AI models that work well with speech and text.

Accomplishments that we're proud of

-I'm proud that I was able to successfully complete the project and bring the full idea to life.

-The add-on now converts voice into text, then turns that text into a well-structured script, and finally generates a realistic voiceover from that script. -All the core AI models and goals of the project were achieved.

-I also managed to connect everything end-to-end — from input to final output — using my own backend and AI integrations.

What we learned

-How to work with AI APIs like Hugging Face and integrate them with backend

-Basics of using Adobe Express Add-on SDK and Code Playground

-How to record and process voice input using MediaRecorder API

What's next for Vocal script ai

-Migrate to paid AI models for improved performance and reliability.

-Refactor the codebase to be more modular, maintainable, and follow clean architecture and SOLID principles.

-Adding Tests

-Enhance the user interface to match Adobe Express design standards.

-Implement persistent storage for user data and history.

-Prepare for a full launch on the Adobe Express Add-on Marketplace.

Built With

Share this project:

Updates