Own-voice/external-speaker detection on Google Speech Commands head-and-torso simulated ATF
98.38Accuracy (Own)Conformer-small
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Conformer-small#Params=307k, Input segment duration=one-second, Evaluation ATF type=head-and-torso ATF, Fine-tuning stage=sequentially fine-tuned up to the head-and-torso ATF stage2026.03 | 98.38 | 91.8 | 96.66 | |
| ResNet#Params=238k, Input segment duration=one-second, Evaluation ATF type=head-and-torso ATF, Fine-tuning stage=sequentially fine-tuned up to the head-and-torso ATF stage2026.03 | 98.24 | 83.18 | 94.3 |