kudu-speech-base
Speech-to-text model for Kudu: NVIDIA Parakeet TDT 0.6B v2.
Benchmarks
Word error rate, lower is better. Published by NVIDIA.
| Dataset | WER |
|---|---|
| Average (8 datasets) | 6.05% |
| LibriSpeech test-clean | 1.69% |
| LibriSpeech test-other | 3.19% |
License
CC-BY-4.0. Model by NVIDIA (Parakeet TDT 0.6B v2).
Model tree for manasdutta04/kudu-speech-base
Base model
nvidia/parakeet-tdt-0.6b-v2