K2-k2/SpeakSync-AI

🎙️ SpeakSync AI

A powerful, real-time voice translation bridge built with Streamlit, OpenAI Whisper, and Google Gemini 2.5 Flash. Speak in your native tongue and let AI handle the transcription, translation, and vocal synthesis instantly.

🌟 Overview

SpeakSync AI is designed to break language barriers by combining state-of-the-art Speech-to-Text (STT), Large Language Models (LLM), and Text-to-Speech (TTS) technologies. Whether you're recording live or uploading a file, the app detects your language and speaks back to you in your chosen target dialect.

🛠️ Key Features

• —Dual Input Modes: Record directly via your microphone or upload audio files (mp3, wav, m4a).
• —Whisper Powered: Uses OpenAI's Whisper (Base) for high-accuracy, multilingual transcription.
• —Gemini 2.5 Flash: Leverages Google's fastest model for nuanced, context-aware translations.
• —Auto-Detect: Not sure what language is being spoken? The AI identifies it for you.
• —Instant Playback: Automatically generates and plays the translated audio using Google Text-to-Speech (gTTS).
• —Smart UI: Features a seamless language swap toggle and persistent session states.

🚀 Deployment on Hugging Face Spaces

1. 1.SDK: Select Streamlit when creating your Space.
2. 2.Secret Keys: Go to Settings > Variables and Secrets*.
3. 3.Add GOOGLE_API_KEY_N with your Gemini API Key from Google AI Studio.
4. 4.Requirements: Ensure your requirements.txt includes:

    streamlit
    openai-whisper
    torch
    google-generativeai
    gTTS

1. 1.System Packages: Create a packages.txt file (required for Whisper/Audio) and add:

    ffmpeg

🏗️ The Tech Stack

• —Frontend: Streamlit
• —STT (Speech-to-Text): OpenAI Whisper
• —Translation Engine: Google Gemini 2.5 Flash
• —TTS (Text-to-Speech): gTTS (Google Text-to-Speech)
• —Processing: Torch (CPU/CUDA optimized)

💻 Local Setup

# Clone the repository
git clone [https://huggingface.co/spaces/K2-k2/SpeakSync-AI](https://huggingface.co/spaces/K2-k2/SpeakSync-AI)

# Install dependencies
pip install -r requirements.txt

# Run the app
streamlit run app.py

Developed by Kesar Agarwal an aspiring AI Engineer focused on creating accessible, agentic AI tools.

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference