🎵 Audio Quality Checker

Deep audio analysis tool for the AI data industry

Features • Quick Start • API Reference • Deployment • Contributing

---

Upload any audio file and get instant quality analysis: signal metrics, language detection, speaker count, speech quality (MOS), noise classification, and compliance checks against industry standards.

✨ Features

Analysis Modes

Mode	Time	What's Included
Quick	~30 sec	Metadata, signal analysis, language detection, VAD, visualizations
Deep	~2-3 min	Everything above + speakers, MOS, noise classification, transcription, emotion

What It Analyzes

📊 Signal Analysis Peak/RMS levels Dynamic range DC offset Signal-to-Noise Ratio (SNR) Clipping detection Silence analysis

🤖 AI Analysis Language detection (Whisper) Speech activity detection (Silero VAD) Speaker diarization (pyannote) Speech quality MOS (NISQA) Noise classification (YAMNet) Transcription (Whisper)

Quality Score

0-100 score with letter grade (A/B/C/D/F) based on weighted metrics:

SNR (25%) + Clipping (15%) + Silence (10%) + Sample Rate (10%) +
Bit Depth (5%) + Dynamic Range (10%) + DC Offset (5%) +
Speech Clarity (10%) + Format Quality (10%) = Total Score

Compliance Profiles

Check audio against industry standards:

Profile	Sample Rate	Bit Depth	SNR	Use Case
General AI Data	≥16 kHz	≥16-bit	≥20 dB	Common requirements
AI Data Platform	≥48 kHz	≥24-bit	≥25 dB	High-quality collection
Crowd Platform	≥16 kHz	≥16-bit	≥15 dB	Crowd-sourced data
Telephony	≥8 kHz	≥16-bit	≥10 dB	Call center audio
Broadcast	≥44.1 kHz	≥16-bit	≥30 dB	Podcast/radio

Supported Formats

WAV MP3 FLAC OGG M4A AAC OPUS WebM AIFF WMA AMR

Limits: Max 1 GB file size • Max 4 hours duration

---

🚀 Quick Start

Requirements

• Python 3.11+
• FFmpeg
• 4GB RAM (8GB recommended for Deep mode)

Installation

# Clone repository
git clone https://github.com/Usergy-ops/audio-quality-checker.git
cd audio-quality-checker

# Setup backend
cd backend
python3 -m venv venv
source venv/bin/activate  # Linux/Mac
pip install -r requirements.txt

# Install AI models (optional, for Deep mode)
pip install torch torchaudio openai-whisper pyannote.audio

# Run server
uvicorn app.main:app --host 0.0.0.0 --port 8000

Open http://localhost:8000 in your browser.

Environment Variables

# Production CORS (comma-separated)
ALLOWED_ORIGINS=https://yourdomain.com

# Rate limiting (default: true)
RATE_LIMIT_ENABLED=true

# HuggingFace token (for speaker diarization)
HF_TOKEN=your_token_here

---

📡 API Reference

Analyze Single File

POST /api/analyze
Content-Type: multipart/form-data

Parameter	Type	Default	Description
file	file	required	Audio file to analyze
mode	string	quick	quick or deep
profile	string	default	Compliance profile
retain	boolean	true	Consent to keep file

Response: Full analysis with quality score, signal metrics, AI analysis, and compliance results.

Batch Analyze

POST /api/analyze-batch
Content-Type: multipart/form-data

Parameter	Type	Default	Description
files	files	required	Up to 20 audio files
mode	string	quick	quick or deep
profile	string	default	Compliance profile

Other Endpoints

Endpoint	Method	Description
/api/profiles	GET	List compliance profiles
/api/limits	GET	Show rate limits
/health	GET	Health check

Rate Limits

Endpoint	Limit
/api/analyze	10/minute per IP
/api/analyze-batch	5/minute per IP
/api/keys/generate	2/hour per IP

---

🌐 Deployment

Production Checklist

• Set ALLOWED_ORIGINS environment variable
• Configure SSL/HTTPS (nginx + Let's Encrypt)
• Set up systemd service
• Configure log rotation
• Set HF_TOKEN for speaker diarization
• Configure firewall

Example nginx Configuration

server {
    listen 443 ssl http2;
    server_name audio.yourdomain.com;
    
    ssl_certificate /etc/letsencrypt/live/audio.yourdomain.com/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/audio.yourdomain.com/privkey.pem;
    
    client_max_body_size 1G;
    
    location / {
        proxy_pass http://127.0.0.1:8000;
        proxy_http_version 1.1;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_read_timeout 600s;
    }
}

Systemd Service

# /etc/systemd/system/audio-checker.service
[Unit]
Description=Audio Quality Checker
After=network.target

[Service]
User=www-data
WorkingDirectory=/opt/audio-quality-checker/backend
Environment="ALLOWED_ORIGINS=https://yourdomain.com"
ExecStart=/opt/audio-quality-checker/backend/venv/bin/uvicorn app.main:app --host 127.0.0.1 --port 8000
Restart=always

[Install]
WantedBy=multi-user.target

---

📁 Project Structure

audio-quality-checker/
├── backend/
│   ├── app/
│   │   ├── main.py              # FastAPI application
│   │   ├── config.py            # Configuration
│   │   ├── routes/
│   │   │   └── analyze.py       # API endpoints
│   │   ├── analyzers/
│   │   │   ├── pipeline.py      # Analysis orchestration
│   │   │   ├── metadata.py      # File info (ffprobe)
│   │   │   ├── signal.py        # Signal analysis
│   │   │   ├── language.py      # Language detection
│   │   │   ├── speakers.py      # Speaker diarization
│   │   │   ├── nisqa.py         # MOS scoring
│   │   │   └── ...
│   │   ├── models/
│   │   │   └── schemas.py       # Pydantic models
│   │   └── utils/
│   │       ├── audio.py         # Audio utilities
│   │       ├── scoring.py       # Quality scoring
│   │       └── profiles.py      # Compliance profiles
│   └── requirements.txt
├── frontend/
│   ├── index.html               # Main UI
│   ├── style.css                # Styles
│   └── app.js                   # Frontend logic
├── tests/
│   └── samples/                 # Test audio files
├── .github/
│   └── ISSUE_TEMPLATE/          # Issue templates
├── README.md
├── CONTRIBUTING.md
├── SECURITY.md
└── .gitignore

---

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md for guidelines.

🔒 Security

Found a vulnerability? Please read SECURITY.md for responsible disclosure.

📄 License

Proprietary — © 2026 UsergyAI. All rights reserved.

---

Built with ❤️ by UsergyAI

Website • Twitter • Email