AudioMOS

AudioMOS is a comprehensive platform for audio processing models, including Text to Speech (TTS), Speech to Text (STT), and Voice Cloning, built on the Akash Network. It offers both API access and a web dashboard for seamless audio processing.

🚀 Features

Text to Speech (TTS)

• Multiple model support
• Language and voice options
• Real-time synthesis

Speech to Text (STT)

• Accurate transcription
• Multiple language support
• File upload capability

Voice Cloning

• Create synthetic voices
• Reference audio support
• High-quality output

Audio Processing

• Noise reduction
• Audio enhancement
• Format conversion

🛠 Installation

Docker Setup (Recommended)

1. Clone the repository

   git clone https://github.com/yourusername/audiomos.git
   cd audiomos
   

2. Launch services

   docker-compose up -d
   

3. Create a virtual environment

   conda create -n audiomos python=3.9
   conda activate audiomos
   

4. Install dependencies

   pip install -r requirements.txt
   

5. Setup backend

   python setup.py
   

6. Start the server

   sh run.sh
   

📜 Example Usage

TTS Example

To convert text to speech:

import requests

response = requests.post(
    "http://localhost:8000/api/tts",
    json={"text": "Hello World", "model": "basic"}
)

# Print the generated audio
with open("output.wav", "wb") as f:
    f.write(response.content)

STT Example

To transcribe speech to text:

import requests

files = {"audio": open("sample.wav", "rb")}
response = requests.post(
    "http://localhost:8000/api/stt",
    files=files
)

# Print the transcribed text
print(response.json()["text"])

🔧 Tech Stack

Backend

• FastAPI: High-performance API framework
• PyTorch: Deep learning framework for model training and inference
• Docker: Containerization for easy deployment
• MongoDB: Database for storing data

Frontend

• React: JavaScript library for building user interfaces
• TypeScript: Typed superset of JavaScript
• Tailwind CSS: Utility-first CSS framework for building custom designs

📝 License

MIT License