🛠 Langswap Goes Open Source
The Langswap project has transitioned its video translation pipeline to Open Source. The system provides end-to-end dubbing: separating audio into speech and background, speech recognition via Whisper with refined VAD boundaries, text translation using Gemma-4-E2B, and voice synthesis via OmniVoice with original timbre cloning.
🌍 Moving key localization tools to Open Source lowers the barrier to entry for creating high-quality dubbing and stimulates the development of models for speech duration control.
👤 It is now possible to deploy a powerful video translation system on your own hardware (NVIDIA GPU) without paying per minute for cloud services, while still preserving the speaker's unique voice.
Source 1: https://github.com/langswap-app/langswap Source 2: https://langswap.app/