Whisper ASR Model for Hindi
The Whisper ASR model for Hindi is an advanced Automatic Speech Recognition (ASR) model developed for accurate and efficient transcription of the Hindi language. With a focus on leveraging cutting-edge technology and innovative methodologies, the Whisper ASR model demonstrates exceptional performance metrics.
Performance Metrics
- Word Error Rate (WER): 13.9913
- Character Error Rate (CER): 5.8844
Overview
The Whisper ASR model has been fine-tuned to accurately transcribe speech in the Hindi language, facilitating seamless communication and accessibility for the Hindi-speaking community. Its performance metrics signify its ability to bridge the gap between human speech and machine comprehension, contributing to the evolution of speech recognition technology. This project is developed under the mentorship of the Kagglex Mentorship Program. The project space can be accessed here, where you can find detailed documentation and resources related to the development and implementation of the Whisper ASR model.
- Downloads last month
- 4