MUMBAI, India, Jan. 23 -- Intellectual Property India has published a patent application (202541124619 A) filed by Gnani Innovations Private Limited, Bengaluru, Karnataka, on Dec. 9, 2025, for 'system and method for generating natural-pronunciation phoneme sequences from code-mixed text.'
Inventor(s) include Thoshith S; and Subhadeep Guchhait.
The application for the patent was published on Jan. 23, under issue no. 04/2026.
According to the abstract released by the Intellectual Property India: "The present disclosure relates to a system (100) and a method for generating natural-pronunciation phoneme sequences from code-mixed text. The system (100) includes a text input module (102) to receive a code-mixed text input including words from multiple natural languages, a language detection module (104) to determine, for each word or sub-word, an origin language using neighboring token information, and a normalization module (106) to apply language-specific text normalization based on the determined origin language. The system (100) further includes one or more language-specific phonemizer modules (108) and a phoneme sequence merger module (110) to merge per-word or sub-word phoneme sequences into a unified phoneme sequence in a shared phoneme space. A text encoder (112) may consume the unified phoneme sequence to generate encoded representations for text-to-speech synthesis or voice cloning."
Disclaimer: Curated by HT Syndication.
