Quick Start

Deploy Smallest Self-Host Speech-to-Text with Docker Compose in under 15 minutes

Overview

This guide walks you through deploying Smallest Self-Host using Docker Compose. You’ll have a fully functional speech-to-text service running in under 15 minutes.

Ensure you’ve completed all prerequisites before starting this guide.

Step 1: Create Project Directory

Create a directory for your deployment:

mkdir -p ~/smallest-self-host
cd ~/smallest-self-host

Step 2: Login to Container Registry

Authenticate with the Smallest container registry using credentials provided by support:

docker login quay.io

Enter your username and password when prompted.

Save your credentials securely. You’ll need them if you restart or redeploy the containers.

Step 3: Create Environment File

Create a .env file with your license key:

cat > .env << 'EOF'
LICENSE_KEY=your-license-key-here
EOF

Replace your-license-key-here with the actual license key provided by Smallest.ai.

Never commit your .env file to version control. Add it to .gitignore if using git.

Step 4: Create Docker Compose File

Lightning ASR (Standard)

Best for: Fast inference, real-time applications

Create a docker-compose.yml file:

docker-compose.yml

version: "3.8"

services:
  lightning-asr:
    image: quay.io/smallestinc/lightning-asr:latest
    ports:
      - "2233:2233"
    environment:
      - MODEL_URL=${MODEL_URL}
      - LICENSE_KEY=${LICENSE_KEY}
      - REDIS_URL=redis://redis:6379
      - PORT=2233
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    restart: unless-stopped
    networks:
      - smallest-network

  api-server:
    image: quay.io/smallestinc/self-hosted-api-server:latest
    container_name: api-server
    environment:
      - LICENSE_KEY=${LICENSE_KEY}
      - LIGHTNING_ASR_BASE_URL=http://lightning-asr:2233
      - API_BASE_URL=http://license-proxy:3369
    ports:
      - "7100:7100"
    networks:
      - smallest-network
    restart: unless-stopped
    depends_on:
      - lightning-asr
      - license-proxy

  license-proxy:
    image: quay.io/smallestinc/license-proxy:latest
    container_name: license-proxy
    environment:
      - LICENSE_KEY=${LICENSE_KEY}
    networks:
      - smallest-network
    restart: unless-stopped

  redis:
    image: redis:7-alpine
    ports:
      - "6379:6379"
    networks:
      - smallest-network
    restart: unless-stopped
    command: redis-server --appendonly yes
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 5s
      timeout: 3s
      retries: 5

networks:
  smallest-network:
    driver: bridge
    name: smallest-network

Step 5: Additional Configuration for Lightning ASR

Lightning ASR

Add the model URL to your .env file (required for Lightning ASR):

echo "MODEL_URL=your-model-url-here" >> .env

The MODEL_URL is provided by Smallest.ai support.

Step 6: Start Services

Launch all services with Docker Compose:

docker compose up -d

Step 7: Monitor Startup

Watch the logs to monitor startup progress:

docker compose logs -f

Look for these success indicators:

1. Redis Ready

   redis-1  | Ready to accept connections
   
2. License Proxy Ready

   license-proxy  | License validated successfully
   license-proxy  | Server listening on port 3369
   
3. Model Service Ready

   Lightning ASR:

   lightning-asr-1  | Model loaded successfully
   lightning-asr-1  | Server ready on port 2233
   
4. API Server Ready

   api-server  | Connected to Lightning ASR
   api-server  | API server listening on port 7100
   

Common Startup Issues

GPU Not Found

Error: could not select device driver "nvidia"

Solution:

sudo systemctl restart docker
docker run --rm --gpus all nvidia/cuda:11.8.0-base-ubuntu22.04 nvidia-smi

If this fails, reinstall NVIDIA Container Toolkit.

License Validation Failed

Error: License validation failed

Solution:

• Verify LICENSE_KEY in .env is correct
• Check internet connectivity
• Ensure firewall allows HTTPS to api.smallest.ai

Model Download Failed

Error: Failed to download model

Solution:

• Verify MODEL_URL in .env is correct
• Check disk space: df -h
• Check internet connectivity

Port Already in Use

Error: port is already allocated

Solution: Check what’s using the port:

sudo lsof -i :7100

Either stop the conflicting service or change the port in docker-compose.yml

Managing Your Deployment

Stop Services

docker compose stop

Restart Services

docker compose restart

View Logs

docker compose logs -f [service-name]

Examples:

docker compose logs -f api-server
docker compose logs -f lightning-asr

Update Images

Pull latest images and restart:

docker compose pull
docker compose up -d

Remove Deployment

Stop and remove all containers:

docker compose down

Remove containers and volumes (including downloaded models):

docker compose down -v

Using -v flag will delete all data including downloaded models. They will need to be re-downloaded on next startup.

What’s Next?

STT Configuration

Customize your deployment with advanced configuration options

STT Services Overview

Learn about each service component in detail

STT Troubleshooting

Debug common issues and optimize performance

API Reference

Integrate with your applications using the API