Speech Recognition On Mediaspeech
评估指标
WER for Arabic
WER for French
WER for Spanish
WER for Turkish
评测结果
各个模型在此基准测试上的表现结果
模型名称 | WER for Arabic | WER for French | WER for Spanish | WER for Turkish | Paper Title | Repository |
|---|---|---|---|---|---|---|
| Wit | 0.2333 | 0.1759 | 0.0879 | 0.0768 | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| Silero | - | - | 0.3070 | - | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| Quartznet | 0.1300 | 0.1915 | 0.1826 | 0.1422 | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| Azure | 0.3016 | 0.1683 | 0.1296 | 0.2296 | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| VOSK | 0.3085 | 0.2111 | 0.1970 | 0.3050 | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| 0.4464 | 0.2385 | 0.2176 | 0.2707 | MediaSpeech: Multilanguage ASR Benchmark and Dataset | ||
| Deepspeech | - | 0.4741 | 0.4236 | - | MediaSpeech: Multilanguage ASR Benchmark and Dataset | |
| wav2vec | 0.9596 | 0.3113 | 0.2469 | 0.5812 | MediaSpeech: Multilanguage ASR Benchmark and Dataset |
0 of 8 row(s) selected.