Keyword Spotting On Hey Siri
Metrics
Error Rate
Results
Performance results of various models on this benchmark
Model Name | Error Rate | Paper Title | Repository |
|---|---|---|---|
| End-to-end DNN-HMM | 1.7% | HEiMDaL: Highly Efficient Method for Detection and Localization of wake-words | - |
| End-to-end DNN-HMM | 1.7% | Optimize what matters: Training DNN-HMM Keyword Spotting Model Using End Metric | - |
| Stacked 1D CNN | 1.99% | HEiMDaL: Highly Efficient Method for Detection and Localization of wake-words | - |
| HEiMDaL | 0.45% | HEiMDaL: Highly Efficient Method for Detection and Localization of wake-words | - |
0 of 4 row(s) selected.