A Whisper model fine-tuned with GRPO (Group Relative Policy Optimization) using a blended error-rate reward. Trained with whisper-rl.

Fine-tuned on fixie-ai/common_voice_17_0 — en, streamed and decoded on the fly from a 1024-clip slice. Each clip's language is pinned from its Common Voice locale during training.

Validation WER and CER at the best checkpoint, per Common Voice locale (also in the Evaluation Results metadata above):

Instead of cross-entropy against a single reference, for each audio clip the policy samples a group of num_generations transcriptions, scores each by a negated blend of word error rate, character error rate, and length / repetition penalties, and is nudged toward the better candidates with a clipped policy-gradient objective regularized by a per-token KL penalty to the frozen base model. Advantages are the group-relative, standardized rewards (A = (r - mean) / (std + eps)), so no value network is needed. The clip's language is pinned from its Common Voice locale, and the policy's own greedy transcriptions are scored as validation WER and CER.

Pulled from the Weights & Biases run (static snapshot):

A proof-of-concept GRPO recipe, not a tuned production system. WER and CER are reported on a held-out Common Voice validation slice after text normalization; real-world performance varies by domain, accent, language, and audio quality.