The IndicAudioDialog Sample Dataset is a multilingual, multichannel, source-separated, conversational speech dataset. It features human-voiced recordings of dialogues translated into 9 Indian languages (Hindi, Tamil, Telugu, Punjabi, Malayalam, Kannada, Bengali, Gujarati, and Marathi) using GPT-4.1. The dataset contains over 30 hours of high-quality audio, recorded by native… See the full description on the dataset page: https://huggingface.co/datasets/snorbyte/indic-audio-dialog-sample.
