indicvoices-cleaned

# IndicVoices Cleaned **IndicVoices Cleaned** is a curated repository containing cleaned transcripts derived from the IndicVoices dataset. We start with normalized verbatim text from IndicVoices and further process it using Google's Gemini model to produce grammatically correct and fluent sentences. These cleaned sentences are intended for use in creating high-quality datasets for: - Machine translation - Transliteration - Other multilingual NLP tasks ## Supported Languages ### Primary Languages We offer robust support for the following 15 major Indian languages: > Assamese, Bengali, Gujarati, Hindi, Kannada, Maithili, Malayalam, Marathi, Nepali, Odia, Punjabi, Sanskrit, Tamil, Telugu, Urdu ### Low-Resource Languages We also provide preliminary support for the following 7 low-resource Indian languages: > Bodo, Dogri, Kashmiri, Konkani, Manipuri, Santali, Sindhi