Parakeet Dictation Cleanup Eval

Parakeet Dictation Cleanup Eval Status (2026-10-16): placeholder. No clips recorded yet. Evaluation only. Never train on this set. Planned contents: 40 spontaneous dictations by one speaker, recorded from fictional scenario cards (recording/scenarios.json in https://github.com/danielrosehill/ASR-Post-Processor-Model), transcribed by stock nvidia/parakeet-tdt-0.6b-v2, and hand-cleaned at both control settings ([Layers: 1] and [Layers: 1,2]). The cast and details on the cards are… See the full description on the dataset page: https://huggingface.co/datasets/danielrosehill/Parakeet-Dictation-Cleanup-Eval.