Model card
wav2vec 2.0 Large (960h).
Meta AIopen-source317M paramsCNN feature encoder + Transformer
wav2vec 2.0 Large fine-tuned on LibriSpeech 960h. Self-supervised pretraining on LibriLight 60k.
§ 02 · Benchmarks
Every benchmark wav2vec 2.0 Large (960h) has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | Common Voice | Speech · Speech Recognition | wer | 10.5% | #2 | 2020-06-20 | source ↗ |
| 02 | LibriSpeech | Speech · Speech Recognition | wer-test-other | 3.3% | #2 | 2020-06-20 | source ↗ |
| 03 | VoxPopuli | Speech · Speech Recognition | wer | 30.1% | #2 | — | source ↗ |
| 04 | AMI-IHM | Speech · Speech Recognition | wer | 42.7% | #4 | — | source ↗ |
| 05 | Earnings-22 | Speech · Speech Recognition | wer | 43.8% | #4 | — | source ↗ |
| 06 | GigaSpeech | Speech · Speech Recognition | wer | 27.7% | #4 | — | source ↗ |
| 07 | Open ASR Leaderboard | Speech · Speech Recognition | wer | 26.8% | #4 | — | source ↗ |
| 08 | SPGISpeech | Speech · Speech Recognition | wer | 22.8% | #4 | — | source ↗ |
| 09 | TED-LIUM | Speech · Speech Recognition | wer | 18.9% | #4 | — | source ↗ |
| 10 | LibriSpeech | Speech · Speech Recognition | wer-test-clean | 1.8% | #5 | 2020-06-20 | source ↗ |
| 11 | Open ASR Leaderboard | Speech · Speech Recognition | rtfx | 516.58 | #18 | — | source ↗ |
| 12 | LibriSpeech | Speech · Speech Recognition | wer | 15.5% | #90 | — | source ↗ |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area
Where wav2vec 2.0 Large (960h) actually performs.
§ 04 · Papers
2 papers with results for wav2vec 2.0 Large (960h).
- 2020-06-20· Speech· 3 results
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
- 2020-06-20· 9 results
wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
§ 05 · Related models
Other Meta AI models scored on Codesota.
GENRE
1 result · 1 SOTA
SeamlessM4T v2 Large
2.3B params · 1 result · 1 SOTA
HuBERT Large (LS-960)
317M params · 2 results
DINOv2 (ViT-g) + Linear
Unknown params · 1 result
Fairseq S2T (MuST-C)
~150M params · 1 result
Mask2Former (Swin-L)
Unknown params · 1 result
MusicGen Large
3.3B params · 1 result
Voicebox
330M params · 1 result
§ 06 · Sources & freshness
Where these numbers come from.
pwc-dump
9
results
arxiv
3
results
3 of 12 rows marked verified.