When Synthetic Speech Is All You Have: Better Call GRPO
LLM-based ASR adapted to regulated domains such as banking is bottlenecked by privacy: real speech is costly and legally constrained to collect, making synthetic text-to-speech (TTS) an attractive substitute. Yet synthetic speech stays acoustically mismatched with real recordings, and work on this gap has stayed within supervised fine-tuning (SFT). We instead turn to reinforcement learning, and show that Group Relative Policy Optimization (GRPO) extracts far more from the same synthetic speech than SFT. Synthetic-only adaptation of the model with GRPO, a critic-free method rewarding low-WER hy
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorShashi Kumar →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorYanis Labrak →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorHasindri Watawana →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorSergio Burdisso →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorEsaú Villatoro-Tello →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorKadri Hacioğlu →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorPetr Motlicek →
When Synthetic Speech Is All You Have: Better Call GRPO
- Linked via arxiv authorAndreas Stolcke →
When Synthetic Speech Is All You Have: Better Call GRPO