The latest articles on DEV Community by prakashkumarmahato807 (@prakashkumarmahato807). https://dev.to/prakashkumarmahato807 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3878605%2F8b1fea12-7d68-44b2-9123-13c851620d53.png https://dev.to/prakashkumarmahato807 en prakashkumarmahato807 Tue, 14 Apr 2026 12:31:35 +0000 https://dev.to/prakashkumarmahato807/building-a-voice-controlled-ai-agent-with-groq-and-streamlit-28i0 https://dev.to/prakashkumarmahato807/building-a-voice-controlled-ai-agent-with-groq-and-streamlit-28i0 <h2> Introduction </h2> <p>I built a Voice Controlled AI Agent for Mem0 internship <br> assignment. It accepts audio input, detects user intent, <br> and executes actions automatically.</p> <h2> How It Works </h2> <ol> <li>User uploads an audio file</li> <li>Groq Whisper converts audio to text</li> <li>LLaMA 3.3 70B detects the intent</li> <li>Tools execute the action</li> <li>Results shown in Streamlit UI</li> </ol> <h2> Architecture </h2> <p>Audio Input<br> ↓<br> Groq Whisper Large V3 (Speech to Text)<br> ↓<br> LLaMA 3.3 70B (Intent Detection)<br> ↓<br> Tool Execution<br> ↓<br> Streamlit UI</p> <h2> Models Used </h2> <p><strong>Speech to Text:</strong> Groq Whisper Large V3</p> <ul> <li>Fast and accurate transcription</li> <li>Supports WAV and MP3 formats</li> <li>Free API available on Groq</li> </ul> <p><strong>Intent Detection:</strong> LLaMA 3.3 70B Versatile</p> <ul> <li>Classifies user intent accurately</li> <li>Fast response time</li> <li>Free on Groq API</li> </ul> <h2> Supported Intents </h2> <ul> <li>create_file: Creates a new file automatically</li> <li>write_code: Generates and saves Python code</li> <li>summarize: Summarizes given text</li> <li>general_chat: General conversation</li> </ul> <h2> Tech Stack </h2> <ul> <li>Python 3.x</li> <li>Streamlit for UI</li> <li>Groq API for STT and LLM</li> <li>Python-dotenv for API key management</li> </ul> <h2> Challenges I Faced </h2> <p><strong>Challenge 1: Model Decommissioned</strong></p> <ul> <li>llama3-8b-8192 model stopped working</li> <li>Solution: Switched to llama-3.3-70b-versatile</li> </ul> <p><strong>Challenge 2: Python PATH Issue on Windows</strong></p> <ul> <li>python command not recognized</li> <li>Solution: Used py -m pip install commands</li> </ul> <p><strong>Challenge 3: GitHub Authentication Error</strong></p> <ul> <li>Permission denied during git push</li> <li>Solution: Used Personal Access Token</li> </ul> <h2> How to Run This Project </h2> <p>Step 1: Clone the repository<br> Step 2: Install dependencies<br> pip install streamlit groq python-dotenv</p> <p>Step 3: Create .env file and add your key<br> GROQ_API_KEY=your_key_here</p> <p>Step 4: Run the app<br> streamlit run app.py</p> <h2> GitHub Repository </h2> <p><a href="https://github.com/prakashkumarmahato807/voice-agent" rel="noopener noreferrer">https://github.com/prakashkumarmahato807/voice-agent</a></p> <h2> Conclusion </h2> <p>Successfully built a working voice agent that processes <br> audio commands and executes actions automatically. <br> This project taught me about speech to text, <br> intent detection, and AI powered web applications.</p> agents ai llm showdev