Ambuj Mehrish

deep reinforcement learning (1) domain adaptation (1) temporal-difference learning (1) speaker diarization (1) low-resource language (1) diffusion model (1) tree search (1) reward model (1) parameter-efficient fine-tuning (1) consistency model (1) best-of-n sampling (1) inference-time alignment (1) parameter-efficient adaptation (1) text-to-speech synthesis (1) error correction (1) speaker adaptation (1) voice cloning (1) speaker representation (1) speaker segmentation (1) speech annotation (1)