Primepluto/

hinglish-whisper-benchmark

A structured evaluation dataset for testing Whisper on Indian-context audio: code-mixed Hinglish, pure Hindi, Indian English, proper nouns, addresses, and call-center phrases.

	Dataset




	

Category Clips Description

pure_hindi 3 Standard Hindi sentences

pure_english_indian 3 English with Indian vocabulary (UPI, EMI, HDFC)

hinglish 5 Romanized code-mixed Hindi-English

indian_names 3 Indian proper nouns (names, cities)

address_pin 3… See the full description on the dataset page: https://huggingface.co/datasets/Primepluto/hinglish-whisper-benchmark.