MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202511069879 A) filed by Pranet Khetan, Gurugram, Haryana, on July 22, 2025, for 'an end-to-end ai-based automatic speech recognition system for dysarthric speech in low-resource languages.'
Inventor(s) include Pranet Khetan.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "Present disclosure describes techniques for automatic speech recognition (ASR) in at least one low-resource language. The techniques include the step of receiving, from a plurality of users, a plurality of audio samples captured through at least one microphone, wherein the plurality of users suffer from dysarthria and applying a two-step data augmentation procedure including a combination of phase vocoding for speech variation and synthetic sentence generation on the plurality of audio samples to generate labelled audio dataset. The techniques include the step of training the AI model to learns the relationship between sequences of audio spectrogram features and labelled text, based on the labelled audio dataset. The training is performed using a transformer-based model corresponding to at least one low-resource language. The technique finally includes deploying the trained AI model into at least one device for predicting speech of at least one user."
Disclaimer: Curated by HT Syndication.
