Intended Use Cases
ML-SUPERB leaderboard SOTA (tops MMS, w2v-BERT 2.0, XLS-R); SUPERB 4 tasks SOTA
Open — attribution required; commercial use allowed.
Every column the registry tracks. Values marked derivedwere filled by a rule or read from the source's text rather than stated in its metadata; the rule or the quoted sentence is shown under the value.
This is the same multilingual corpus, sliced for other languages the platform indexes it under.
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
Relevant to aao: HuggingFace language filter shows one aao model, espnet/xeus; model covers 4,024 languages and can be fine-tuned for Algerian Saharan Arabic…
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
General multilingual speech encoder that can be used for Hejazi Arabic speech work if acw data is available. Not an acw-specific fine-tuned model.
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
automatic-speech-recognition model · espnet · espnet
0.6B-parameter Afrocentric MT model. Supports Afar (aar) among 46 languages in 150+ pairs. Fine-tuned versions: toucan-1.2B and toucan-3.7B also available. T…
1.2B-parameter fine-tuned MT variant of Cheetah-1.2B. Significantly outperforms other MT models for African languages. Afar (aar) supported. Also available: …
1.2B-parameter Afrocentric sequence-to-sequence pretrained LM covering 517 African languages. Afar (aar) included. Trained on 42G corpus from 50 of 54 Africa…