Setup & Optimization

Self-Hosted Remote Services Setup Guide

Run server-grade OpenAI Whisper speech recognition and LibreTranslate on your local network. Perfect for budget hardware, tablets, or anyone who wants maximum transcription accuracy with complete privacy.

Video Tutorial: Self-Hosted Whisper & LibreTranslate

Watch this step-by-step video guide showing how to start the Docker containers, find your local IP, and connect LiveCaptionN:

System Requirements & Architecture

LiveCaptionN gives you the choice between two transcription and translation pipelines:

Requirement Tier	Specs (RAM / CPU / Storage)	Recommended Mode
Passed Fully Supported	≥ 4 GB RAM, 64-bit 6+ cores, ≥ 1.5 GB free storage	On-Device (Vosk & ML Kit) or Remote
Borderline Low Resource	2.5 – 3.8 GB RAM (3 GB devices, e.g. Kindle Fire HD 8), 32-bit or 4 cores	Remote Whisper recommended (small on-device allowed with warning)
Unsupported Below Minimum	< 2.5 GB RAM, < 4 CPU cores, or < 350 MB free space	Remote Whisper & LibreTranslate (Enforced)

Why use self-hosted services? On-device neural models require heavy CPU and RAM. By running Whisper ASR and LibreTranslate in Docker on your laptop or home desktop, your phone uses almost no battery or RAM while receiving instant, high-accuracy captions over your local Wi-Fi.

1 Install Docker Desktop

If you do not already have Docker installed on your PC or Mac:

Windows: Download Docker Desktop for Windows (requires WSL 2 enabled).
Mac: Download Docker Desktop for Mac (Apple Silicon or Intel).
Linux: Install Docker Engine via your package manager (e.g. sudo apt install docker.io).

2 Start the Whisper Speech-to-Text Server

Run the high-performance Whisper ASR container with a single command in your terminal or PowerShell:

docker run -d -p 9000:9000 -e ASR_MODEL=base -e ASR_ENGINE=openai_whisper onerahmet/openai-whisper-asr-webservice

Model options: You can replace ASR_MODEL=base with small, medium, or large-v3 if your computer has a dedicated GPU.
Port: The service listens on port 9000.
Verify: Open your browser to http://localhost:9000/docs to see the Swagger API documentation.

3 Start the LibreTranslate Server

Run the self-hosted LibreTranslate container in your terminal:

docker run -d -p 5000:5000 libretranslate/libretranslate

Languages: Automatically downloads official language translation packages on first boot.
Port: The service listens on port 5000.
Verify: Open your browser to http://localhost:5000 to verify the translation web interface.

4 Find Your Computer's Local IP Address

Your phone connects to your computer over your local home Wi-Fi. Find your computer's local IP:

Windows: Open PowerShell and run ipconfig. Look for IPv4 Address under your Wi-Fi adapter (e.g., 192.168.1.50 or 10.0.0.74).
macOS: Open System Settings > Wi-Fi > Details, or run ipconfig getifaddr en0 in Terminal.
Linux: Run ip a or hostname -I.

5 Configure LiveCaptionN on Your Android Device

In the LiveCaptionN app:

Make sure your Android device is connected to the same Wi-Fi network as your computer.
Open LiveCaptionN and scroll down to Server Settings.
Under Speech-to-Text URL (Whisper), enter:
http://<YOUR_LOCAL_IP>:9000/asr?output=json
Example: http://192.168.1.50:9000/asr?output=json
Under LibreTranslate URL, enter:
http://<YOUR_LOCAL_IP>:5000
Example: http://192.168.1.50:5000
Tap Save and Refresh. The app will fetch the available languages directly from your server!

Troubleshooting & Tips

Firewall blocking: On Windows, if your phone cannot reach your PC, ensure Windows Defender Firewall allows incoming connections on private networks for Docker or ports 9000 and 5000.
Check from phone browser: Test by opening http://<YOUR_LOCAL_IP>:5000 in Chrome on your phone. If it opens, LiveCaptionN will connect instantly.
Auto-start on boot: Add --restart unless-stopped to both docker commands so your transcription services start automatically whenever your computer boots.