From the model card
Fields below are copied from the tags and counters on the HuggingFace repository nvidia/speakerverification_en_titanet_large at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- nvidia
- Library
- NeMo
- Framework tags
- PyTorch
- License tag
cc-by-4.0— read the license file in the repo before relying on it- Language tags
- English (en)
- Datasets declared
- VOXCELEB-1, VOXCELEB-2, FISHER, switchboard, librispeech_asr, SRE
- Downloads (HF counter at last fetch)
- 350,886
- Likes (HF counter at last fetch)
- 122
- Model card
- https://huggingface.co/nvidia/speakerverification_en_titanet_large
Use cases
- Speaker verification in voice authentication systems
- Speaker diarisation (who spoke when) in meeting transcription pipelines
- Building speaker-aware conversation indexing systems
- Voice identity verification for call centre access control
- Research on speaker representation learning and embedding quality
Pros
- NeMo-native with published EER benchmarks on VoxCeleb and similar datasets
- Integrates directly with NeMo's diarisation pipeline for complete speaker tracking
- 120 likes with active use in production diarisation systems
- Large variant provides stronger speaker discrimination than base TitaNet
Cons
- NeMo dependency; requires the full NeMo toolkit rather than standard transformers
- English-trained; cross-language speaker embedding quality is not guaranteed
- Speaker verification performance degrades significantly with short utterances under 3 seconds
- No license explicitly stated; check NVIDIA NeMo model terms