NVIDIA stt_de_fastconformer_hybrid_large_pc, original revision f4df00ef37d9b044cb19d824dde17cc65f68630e. https://huggingface.co/nvidia/stt_de_fastconformer_hybrid_large_pc
CC BY 4.0 https://creativecommons.org/licenses/by/4.0/; license text and upstream card retained here.
INT8 ONNX export by csukuangfj / sherpa-onnx, revision df32693bc4328136c455af24035a5ee1ec708df0, https://huggingface.co/csukuangfj/sherpa-onnx-nemo-transducer-stt_de_fastconformer_hybrid_large_pc-int8 . Conversion and quantization from NVIDIA NeMo weights; no weight edits in this QA. No creator endorsement implied. Native sherpa-onnx runtime Apache 2.0; ONNX Runtime MIT and third-party notices retained. This diagnostic is German transcription only, not a translation service.

This optional local speech pack transcribes explicitly selected German dialogue.
Conversion/quantization modifications are credited above; WebCast+ does not claim
author endorsement or automatic mixed-language accuracy. The CC-BY-4.0 legal
code is shipped as CC-BY-4.0-GERMAN-NEMO.txt. Existing Sherpa/ONNX Runtime notices
continue to apply. Attribution is in application licenses, never a video overlay.