Model
SpeakerNet-M model for Speaker Diarization inference
Use the NGC CLI to download:
Copied!
1 Version
1.0.0rc1Selected03/10/2021 3:21 AM UTC24.09 MB Copied!
1.0.0rc1Selected
03/10/2021 3:21 AM UTC24.09 MB
Copied!
Acccuracy
| Key | Value |
|---|---|
| SER on CH109 set | 8.3% |
MODEL
| Key | Value |
|---|---|
| Architecture | QuartzNet |
| inputs | Audio |
| outputs | Speaker Matching Label and Embeddings |