Automatic Speech Recognition on Telephony short (test)
21.24WER (%)ConformerXL-RNNT-PS (Fine-tuned)
Evaluation results
| Method | Links | |
|---|---|---|
| ConformerXL-RNNT-PS (Fine-tuned)Fine-tuning Mixture (Telephony)=0.8, Fine-tuning Mixture (Video)=0.2, Fine-tuning Mixture (NST)=N/A2021.09 | 21.24 | |
| ConformerXL-RNNT-PFine-tuning Mixture (Telephony)=0.8, Fine-tuning Mixture (Video)=0.2, Fine-tuning Mixture (NST)=N/A2021.09 | 22.24 | |
| Streaming Model (Fine-tuned)Fine-tuning Mixture (Telephony)=1.0, Fine-tuning Mixture (Video)=N/A, Fine-tuning Mixture (NST)=N/A2021.09 | 22.45 | |
| Streaming Model (Fine-tuned)Fine-tuning Mixture (Telephony)=0.8, Fine-tuning Mixture (Video)=0.2, Fine-tuning Mixture (NST)=N/A2021.09 | 22.64 | |
| Student Streaming ModelFine-tuning Mixture (Telephony)=0.8, Fine-tuning Mixture (Video)=0.2, Fine-tuning Mixture (NST)=0.22021.09 | 22.97 | |
| ConformerXL-RNNT-PS (Baseline)Fine-tuning Mixture (Telephony)=N/A, Fine-tuning Mixture (Video)=N/A, Fine-tuning Mixture (NST)=N/A2021.09 | 27.2 | |
| Streaming Model (Baseline)Fine-tuning Mixture (Telephony)=N/A, Fine-tuning Mixture (Video)=N/A, Fine-tuning Mixture (NST)=N/A2021.09 | 33.11 |