IISc Develops Text-to-Speech Models and Speech Corpus for Sourashtra Language
bengaluru: the indian institute of science (iisc), in collaboration with the digital india bhashini division under the ministry of electronics and information technology (meity), has developed and released text-to-speech (tts) models and a speech corpus for sourashtra, an indo-aryan language predominantly spoken by the sourashtra community.the initiative marks the first such release for the language and is an important step towards developing digital speech resources for languages that have limited representation in india's artificial intelligence (ai) ecosystem, iisc said in a statement.developed at iisc's speech and image processing and recognition (spire) lab, the resource comprises four tts models trained using more than 28 hours of gender-balanced speech recordings.the models can convert written sourashtra text into spoken audio, offering male and female voices in both tamil and sourashtra scripts.the underlying spicor tts 3.0 sourashtra corpus has been made freely accessible to researchers, developers and institutions through aikosh, india's national platform for ai datasets and models.the corpus covers diverse subjects, including finance, politics, food, arts, literature, weather and science. it contains sentence-level data that can be used to develop and evaluate speech technologies across different contexts and vocabulary.prasanta kumar ghosh, professor at iisc, said ai could play a significant role in making digital content more accessible to communities whose languages have limited representation in existing technologies.the newly developed models would help convert written sourashtra content into natural-sounding speech across two scripts, he added.according to the 2011 census, around 2.48 lakh people speak sourashtra, iisc said.