Speech Recognition On Switchboard Hub500
Metrics
Percentage error
Results
Performance results of various models on this benchmark
Comparison Table
| Model Name | Percentage error |
|---|---|
| Model 1 | 11 |
| Model 2 | 12.9 |
| Model 3 | 12.9 |
| the-ibm-2015-english-conversational-telephone | 8.0 |
| the-microsoft-2016-conversational-speech | 6.2 |
| Model 6 | 11.5 |
| Model 7 | 18.5 |
| the-ibm-2016-english-conversational-telephone | 6.6 |
| building-dnn-acoustic-models-for-large | 16 |
| Model 10 | 12.9 |
| Model 11 | 10.4 |
| deep-speech-scaling-up-end-to-end-speech | 12.6 |
| single-headed-attention-based-sequence-to | 4.7 |
| building-dnn-acoustic-models-for-large | 15 |
| achieving-human-parity-in-conversational | 6.6 |
| the-microsoft-2016-conversational-speech | 6.3 |
| on-the-limit-of-english-conversational-speech | 4.3 |
| english-conversational-telephone-speech | 5.5 |
| Model 19 | 12.6 |
| the-microsoft-2016-conversational-speech | 6.9 |
| Model 21 | 8.5 |
| deep-speech-scaling-up-end-to-end-speech | 12.6 |
| Model 23 | 9.2 |
| the-ibm-2016-english-conversational-telephone | 6.9 |
| Model 25 | 12.9 |
| Model 26 | 16.1 |
| deep-speech-scaling-up-end-to-end-speech | 20 |
| achieving-human-parity-in-conversational | 5.8 |
| Model 29 | 12.6 |
| very-deep-multilingual-convolutional-neural | 12.2 |