nemo-speech-asr-finetune

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). This skill’s required workflow is training/evaluation over user-supplied ASR JSONL manifests (e.g., /data/test.json in references/data-lhotse.md and references/evaluation-style-contract.md), where the manifest’s free-form text/pred_text strings are read/processed at runtime and can contain outsider-authored transcript content.