🎙️ cf_ai_echomind

Intelligent Voice Journaling with Persistent AI Memory

EchoMind AI is a full-stack, voice-enabled journaling application built entirely on the Cloudflare ecosystem. It allows users to capture spoken thoughts, transcribes them using Whisper, and generates high-reasoning summaries and mood analysis via Llama 3.3—all while maintaining a rolling history of the session through Durable Objects.

---

🚀 Live Demo

Production URL: https://cf-ai-echomind.k-nadarkhani.workers.dev

---

✨ Key Features

• Voice-to-Text Transcription: Native browser audio capture processed through the @cf/openai/whisper model.
• Intelligent Summarization: Deep reasoning and mood detection powered by @cf/meta/llama-3.3-70b-instruct.
• Stateful Coordination: Utilizes Durable Objects to maintain a persistent history of the last 5 entries, providing context for the AI.
• Zero-Latency UI: A lightweight Vanilla JS frontend served directly from the edge via the Worker.

---

🛠️ The Tech Stack

Component	Technology	Role
Orchestration	Cloudflare Workers	Handles routing, UI serving, and AI coordination.
State/Memory	Durable Objects	Manages persistent journal history (SessionObject).
Transcription	OpenAI Whisper	Converts WebM audio blobs to text.
Inference	Llama 3.3 70B	Generates summaries and analyzes sentiment.
Frontend	HTML5 / JS	Captures audio via MediaRecorder API.

---

🧠 How It Works

1. Capture: User records audio; the browser generates a WebM blob.
2. Processing: The Worker receives the blob and runs @cf/openai/whisper to transcribe the speech.
3. Memory: The transcript is sent to a Durable Object to be stored in the user's session history.
4. Analysis: The transcript is passed to Llama 3.3, which returns a concise summary and mood assessment.
5. Output: The final result is rendered in a clean, modern UI card.

---

💻 Local Development

1. Clone & Install:

   git clone [https://github.com/knd8412/cf_ai_echomind.git](https://github.com/knd8412/cf_ai_echomind.git)
   npm install

2. Wrangler Configuration: Ensure your wrangler.toml includes the SESSION_STATE Durable Object binding and the AI binding.
3. Run Locally:

   npx wrangler dev

4. Deploy:

   npx wrangler deploy