wav2vec

Reported on 5 benchmarks across 1 task · 2 papers · 4 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Audio5 results

Speech RecognitiononMediaSpeech
WER for Arabic· 2021-03-30
0.9596
best: 0.13 (Quartznet)
SOTA
MediaSpeech: Multilanguage ASR Benchmark and Dataset arXiv:2103.16193
Speech RecognitiononMediaSpeech
WER for French· 2021-03-30
0.3113
best: 0.1683 (Azure)
SOTA
MediaSpeech: Multilanguage ASR Benchmark and Dataset arXiv:2103.16193
Speech RecognitiononMediaSpeech
WER for Spanish· 2021-03-30
0.2469
best: 0.0879 (Wit)
SOTA
MediaSpeech: Multilanguage ASR Benchmark and Dataset arXiv:2103.16193
Speech RecognitiononMediaSpeech
WER for Turkish· 2021-03-30
0.5812
best: 0.0768 (Wit)
SOTA
MediaSpeech: Multilanguage ASR Benchmark and Dataset arXiv:2103.16193
Speech RecognitiononTIMIT
Percentage error· uses extra data· 2019-04-11
14.7
best: 8.3 (wav2vec 2.0)
wav2vec: Unsupervised Pre-training for Speech Recognition arXiv:1904.05862