Thermal feedback has the potential to enrich immersive interaction, yet its role in material discrimination remains underexplored. We designed and evaluated a conductivity model that simulates transient heating and cooling profiles based on material properties of an object. Thirty-eight participants used a Meta Quest 3 headset and WEART TouchDIVER Pro gloves to classify virtual blocks (metal, glass, wood) under three conditions: visual-thermal congruence, thermal-only, and visual-thermal incongruence. Although the results show accuracy decreased when congruence decreased, the objects were rated consistently following their material conductivity properties (Metal > Glass > Wood), supporting the validity of the conductivity model. Haptic Experience (HX) ratings of realism, involvement, and harmony remained stable across tasks, while sorting difficulty was lowest under congruent and highest under incongruent visuals. Physiological baselines did affect performance. Our findings demonstrate that conductivity-based thermal rendering enables perceptually reliable material differences in VR, informing design and application of thermal haptics.

Lack of large-scale pretraining data for low resource languages from the Indian sub-continent, leads to their underrepresentation in existing massively multilingual models. In this work, we address this gap by proposing a framework to create large raw audio datasets for such under-represented languages by collating publicly accessible audio content. Leveraging this framework, we present MahaDhwani, a corpus comprising 279K hours of raw audio across 22 Indian languages. To test the utility of MahaDhwani, we pretrain a conformer style model, and then further finetune it to build a multilingual ASR model supporting the 22 languages. Using a hybrid multi-softmax decoder, we balance the benefit of shared parameters which enable crosslingual transfer, and the benefit of dedicated capacity for each language. Our evaluations on the IndicVoices benchmark show the benefits of pre-training, particularly in low-resource settings. We will open-source our framework, code and scripts to reproduce the dataset.