Automatic Speech Recognition on LibriSpeech2Mix simulated (eval)
3.56Word Error RateA4 (Table 1)
Evaluation results
| Method | Links | |
|---|---|---|
| A4 (Table 1)Enrollment duration (ED)=5s, proposed_variant=CALM2026.01 | 3.56 | |
| A3 (Table 1)Enrollment duration (ED)=5s, proposed_variant=CALM2026.01 | 3.89 | |
| A2 (Table 1)Enrollment duration (ED)=5s2026.01 | 4.28 | |
| A1 (Table 1)Enrollment duration (ED)=5s2026.01 | 4.58 | |
| CONF-TSASREnrollment duration (ED)=7.5s2026.01 | 6.3 | |
| Transformer SA-ASR2026.01 | 6.4 | |
| Whisper-large-SS-TTI2026.01 | 6.99 |