In this manuscript, we propose a language-specific fine- tuning approach for speaker identification in the complex, multilingual Indian linguistic environment. We fine-tune a TitaNet-L speaker model on a custom 1 8 4-hour corpus of Indian English and Tamil speech, achieving 84.96% Top-1 identification accuracy a 1 4% absolute improvement over the baseline TitaNetL model. A Conformer-CTC ASR model is used for verifying the generated speaker-attributed transcripts in terms of qualitative assessment. The primary contribution of this work is demonstrating that language-specific fine-tuning provides a practical and highly effective path to build accurate speaker identification systems for multilingual Indian environments.