facebook
/

xm_transformer_unity_hk-en

speech-to-speech-translation

Model card Files Files and versions Community

lpw commited on Oct 17, 2022

Commit

433f7f3

·

1 Parent(s): a51c8ed

Update README.md

Files changed (1) hide show

README.md +4 -2

README.md CHANGED Viewed

@@ -16,10 +16,12 @@ datasets:
 ---
 ## xm_transformer_unity_hk-en
-Speech-to-speech translation model from fairseq S2UT:
 - Hokkien-English
-- Trained on MuST-C, TAT and Hokkien dramas
 - Speech synthesis with [facebook/unit_hifigan_mhubert_vp_en_es_fr_it3_400k_layer11_km1000_lj_dur](https://huggingface.co/facebook/unit_hifigan_mhubert_vp_en_es_fr_it3_400k_layer11_km1000_lj_dur)
 ## Usage
 ```python

 ---
 ## xm_transformer_unity_hk-en
+Speech-to-speech translation model with two-pass decoder (UnitY) from fairseq:
 - Hokkien-English
+- Trained with supervised data in TED, drama, [TAT](https://sites.google.com/speech.ntut.edu.tw/fsw/home/tat-corpus) domain, and weakly supervised data in drama domain. See [here](https://research.facebook.com/publications/hokkien-direct-speech-to-speech-translation)
+for training details.
 - Speech synthesis with [facebook/unit_hifigan_mhubert_vp_en_es_fr_it3_400k_layer11_km1000_lj_dur](https://huggingface.co/facebook/unit_hifigan_mhubert_vp_en_es_fr_it3_400k_layer11_km1000_lj_dur)
+- [Examples](https://github.com/facebookresearch/fairseq/tree/ust/examples/hokkien)
 ## Usage
 ```python