MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641110597 A) filed by Vallurupalli Nageswara Rao Vignana Jyothi Institute Of Engineering And Technology on September 15, 2026, for An Ai-Based Multilingual Video Translation System With Emotion-Preserving Speech Synthesis And Synchronized Audio-Visual Rendering.
Inventor includes Mrs. Y. Prasanthi.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: ABSTRACT [0019] A modular video translation pipeline has been developed that converts the spoken content of a video into a target language while maintaining synchronization with the visual track and retaining emotional expressiveness. Audio is first extracted from the uploaded video using FFmpeg and stored in high-fidelity WAV format. The extracted audio is transcribed by OpenAI’s Whisper model, which produces accurate text even under accent and noise variation. The transcript is translated by the Google Translate API into the user-selected target language. The translated text is converted into natural-sounding speech by a pipeline of text-to- speech engines that includes Microsoft Edge TTS, Suno Bark, Coqui XTTS and Piper TTS, optionally enhanced by Sonic Pi for improved intonation. The synthesized audio is then aligned with the original video frames using FFmpeg and Pydub so that lip movements and scene timing remain coherent. A Gradio-based web interface allows non-technical users to upload a video, choose source and target languages, monitor progress and download the finished result. Evaluation demonstrated 91 percent word-level transcription accuracy, natural voice quality and excellent audiovisual synchronization, making the system suitable for educational, entertainment, business and accessibility applications across multiple languages.
Disclaimer: Curated by HT Syndication.
