Schedule-AI Backend is a serverless AI-powered service that helps convert audio to text and extract structured schedule information from it. It uses Cloudflare AI models, Whisper for audio-to-text transcription and LLaMA for extracting structured schedule data from text.

Convert audio files to text

Extract schedule information from text

Structured and reusable API endpoints

TypeScript / Node.js

Cloudflare Workers (serverless deployment)

OpenAI Whisper & LLaMA models

Vitest for testing

HTTP Layer (src/index.js) – Handles incoming requests and routes them to the appropriate services.

Service Layer (src/schedular/) – Orchestrates audio transcription and schedule extraction.

AI Layer (src/llms/) – Contains logic for interacting with AI models (Whisper for audio-to-text, LLaMA for schedule extraction).

Utility Layer (src/utils/) – Provides helper functions like parsing AI responses into structured schedules.

User uploads audio => /audio-to-text endpoint → AI transcription → JSON response

User submits text => /extract-values endpoint → AI extraction → structured schedule returned

Utilities parse the AI response for consistent output

Send audio data to /audio-to-text to get transcription

Send text and timestamp to /extract-values to get structured schedule

MIT © Muhammad Azhari Iqbal