41 One-Command AI Video Translation Pipeline | WSL2 Setup for Indian & International Languages

Original author：Sudheendra S G

Original source：https://www.youtube.com/embed/kwFQq5WqWOQ

00:00 Introduction: Automating the AI Audio Pipeline 01:03 Two-Script Automation: Setup File + Production Runner 02:07 Downloading the Session Files 03:05 Audio Pipeline Folder Structure Explained 04:30 Checking Existing WSL Installation 06:05 Installing Ubuntu in WSL2 07:05 Exporting Ubuntu to D Drive 08:05 Importing Ubuntu Back to Custom Location 09:25 Creating a Custom WSL Shortcut 10:55 Updating Ubuntu Packages 11:35 Running the One-Time WSL2 Setup Script 13:25 Fixing Conda Not Found After Setup 14:43 Hugging Face Model Access Setup 16:30 Authenticating Hugging Face in WSL2 17:35 Overview of the Production Python Scripts 18:35 Running the Full Production Pipeline 19:40 Setting Input Video and Language Parameters 20:25 Adding Reference Audio and Reference Text 21:40 Executing the One-Command Pipeline 22:20 Transcription, Translation, and TTS Execution 23:10 Final Video, Audio, Subtitle, and Output Files 25:05 Conclusion and Next Step: Indic Parler TTS Github Repository https://github.com/ssathvick/indicftts.git For free resources and course materials visit the below patreon page https://www.patreon.com/posts/151488430 In this video, we automate the complete local AI audio translation and dubbing pipeline inside WSL2. Earlier, we completed the pipeline step by step: WSL2 setup, Conda and Mamba installation, PyTorch GPU setup, audio extraction, transcription using Faster-Whisper and Indic Conformer, translation using IndicTrans2, and text-to-speech using IndicF5 and XTTS. But doing all these phases manually every time is not practical for production. So in this session, we convert the entire workflow into an automated system using two main scripts. The first script is a one-time WSL2 machine setup file. It installs Conda, Mamba, FFmpeg, PyTorch, TorchCodec, Hugging Face support, and all required environments for Indian and international language processing. The second script is the production runner. With just an input video, source language, target language, reference voice, and a few parameters, it automatically runs the full workflow: audio extraction, transcription, translation, TTS generation, audio processing, subtitle generation, and final translated video export. This pipeline supports Indian language workflows using IndicTrans2 and IndicF5, and international language workflows using NLLB and XTTS-v2. Topics covered: ✅ One-time WSL2 setup automation ✅ Conda and Mamba environment creation ✅ Faster-Whisper transcription ✅ Indic Conformer for Indian language ASR ✅ IndicTrans2 translation workflow ✅ IndicF5 TTS for Indian languages ✅ XTTS-v2 for international TTS ✅ TorchCodec and FFmpeg fixes ✅ Hugging Face authentication setup ✅ One-command production pipeline ✅ Final dubbed video export This is useful for creators, educators, developers, AI filmmakers, and anyone building a local AI dubbing or translation workflow without depending fully on cloud services. Commands, scripts, and setup files can be adapted for your own local AI production pipeline. #LocalAI #AIDubbing #WSL2 #IndicTrans2 #IndicF5 #XTTS #FasterWhisper #AITranslation #VideoDubbing #OpenSourceAI #AIWorkflow #MachineLearning #TextToSpeech #SpeechToText #FFmpeg