Vividh-ASR: Diagnosing and Fixing Studio-Bias in Whisper for Indic Languages
• 12
AI & ML interests
Speech and NLP systems for Indian legal infrastructure
Recent Activity
View all activity
Papers
SCRIBE: Diagnostic Evaluation and Rich Transcription Models for Indic ASR
Vividh-ASR: A Complexity-Tiered Benchmark and Optimization Dynamics for Robust Indic Speech Recognition
Articles
Accepted to Main Track
Accepted to NLLP Workshop
-
Are LLMs Court-Ready? Evaluating Frontier Models on Indian Legal Reasoning
Paper • 2510.17900 • Published • 1 -
adalat-ai/indian-legal-exam-benchmark
Viewer • Updated • 6.22k • 28 -
CourtNav: Voice-Guided, Anchor-Accurate Navigation of Long Legal Documents in Courtrooms
Paper • 2601.05255 • Published • 1 -
adalat-ai/Indian-Legal-Retrieval-Generation
Viewer • Updated • 21 • 15
Malayalam benchmark and models
-
adalat-ai/vividh-test-malayalam
Viewer • Updated • 15.7k • 137 • 1 -
adalat-ai/whisper-medium-ml-rmft
Automatic Speech Recognition • 0.8B • Updated • 18 • 1 -
adalat-ai/ct2-whisper-medium-ml-rmft
Automatic Speech Recognition • Updated • 33 -
adalat-ai/whisper-medium-ml-high-lr
Automatic Speech Recognition • 0.8B • Updated • 1
-
adalat-ai/whisper-small-ml-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 21 -
adalat-ai/whisper-small-kn-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 18 -
adalat-ai/whisper-small-hi-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 18 -
SCRIBE: Diagnostic Evaluation and Rich Transcription Models for Indic ASR
Paper • 2605.20712 • Published • 4
Vividh ASR Model Checkpoints and Benchmark datasets
Hindi benchmark and models
-
adalat-ai/vividh-test-hindi
Viewer • Updated • 22k • 136 • 1 -
adalat-ai/whisper-medium-hi-rmft
Automatic Speech Recognition • 0.8B • Updated • 70 -
adalat-ai/whisper-medium-hi-high-lr
Automatic Speech Recognition • 0.8B • Updated • 50 -
adalat-ai/whisper-small-hi-rmft
Automatic Speech Recognition • 0.2B • Updated • 1
CTranslate2 ports of IndicTrans2-RoPE collection
Accepted to Main Track
Vividh ASR Model Checkpoints and Benchmark datasets
Accepted to NLLP Workshop
-
Are LLMs Court-Ready? Evaluating Frontier Models on Indian Legal Reasoning
Paper • 2510.17900 • Published • 1 -
adalat-ai/indian-legal-exam-benchmark
Viewer • Updated • 6.22k • 28 -
CourtNav: Voice-Guided, Anchor-Accurate Navigation of Long Legal Documents in Courtrooms
Paper • 2601.05255 • Published • 1 -
adalat-ai/Indian-Legal-Retrieval-Generation
Viewer • Updated • 21 • 15
Hindi benchmark and models
-
adalat-ai/vividh-test-hindi
Viewer • Updated • 22k • 136 • 1 -
adalat-ai/whisper-medium-hi-rmft
Automatic Speech Recognition • 0.8B • Updated • 70 -
adalat-ai/whisper-medium-hi-high-lr
Automatic Speech Recognition • 0.8B • Updated • 50 -
adalat-ai/whisper-small-hi-rmft
Automatic Speech Recognition • 0.2B • Updated • 1
Malayalam benchmark and models
-
adalat-ai/vividh-test-malayalam
Viewer • Updated • 15.7k • 137 • 1 -
adalat-ai/whisper-medium-ml-rmft
Automatic Speech Recognition • 0.8B • Updated • 18 • 1 -
adalat-ai/ct2-whisper-medium-ml-rmft
Automatic Speech Recognition • Updated • 33 -
adalat-ai/whisper-medium-ml-high-lr
Automatic Speech Recognition • 0.8B • Updated • 1
CTranslate2 ports of IndicTrans2-RoPE collection
-
adalat-ai/whisper-small-ml-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 21 -
adalat-ai/whisper-small-kn-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 18 -
adalat-ai/whisper-small-hi-curated-reverse-mft-1-1-1
Automatic Speech Recognition • 0.2B • Updated • 18 -
SCRIBE: Diagnostic Evaluation and Rich Transcription Models for Indic ASR
Paper • 2605.20712 • Published • 4