Update README.md

README.md

@@ -16,13 +16,12 @@ widget:
 - example_title: Common Voice sample 1
   src: https://huggingface.co/facebook/xm_transformer_600m-es_en-multi_domain/resolve/main/common_voice_es_19966634.flac
 ---
-
 ## xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022
 
-speech-to-speech translation model from fairseq S2UT (paper/code):
+Speech-to-speech translation model from fairseq S2UT ([paper](https://arxiv.org/abs/2204.02967)/[code](https://github.com/facebookresearch/fairseq/blob/main/examples/speech_to_speech/docs/enhanced_direct_s2st_discrete_units.md)):
 - Spanish-English
 - Trained on 
-- Speech synthesis with facebook/unit_hifigan_mhubert_vp_en_es_fr_it3_400k_layer11_km1000_lj_dur
+- 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
@@ -35,20 +34,21 @@ from fairseq import hub_utils
 from fairseq.checkpoint_utils import load_model_ensemble_and_task_from_hf_hub
 from fairseq.models.speech_to_text.hub_interface import S2THubInterface
 from fairseq.models.text_to_speech import CodeHiFiGANVocoder
-from fairseq.models.text_to_speech.hub_interface import (
-    TTSHubInterface,
-    VocoderHubInterface,
-)
+from fairseq.models.text_to_speech.hub_interface import VocoderHubInterface
+
 from huggingface_hub import snapshot_download
 import torchaudio
 
+cache_dir = os.getenv("HUGGINGFACE_HUB_CACHE")
 
 models, cfg, task = load_model_ensemble_and_task_from_hf_hub(
     "facebook/xm_transformer_s2ut_800m-es-en-st-asr-bt_h1_2022",
-    arg_overrides={"config_yaml": "config.yaml"},
+    arg_overrides={"config_yaml": "config.yaml", "task": "speech_to_text"},
+    cache_dir=cache_dir,
 )
-model = models[0]
-generator = task.build_generator(model, cfg)
+model = models[0].cpu()
+cfg["task"].cpu = True
+generator = task.build_generator([model], cfg)
 
 
 # requires 16000Hz mono channel audio
@@ -57,9 +57,7 @@ audio, _ = torchaudio.load("/path/to/an/audio/file")
 sample = S2THubInterface.get_model_input(task, audio)
 unit = S2THubInterface.get_prediction(task, model, generator, sample)
 
-# speech synthesis
-cache_dir = os.getenv("HUGGINGFACE_HUB_CACHE")
-            
+# speech synthesis           
 library_name = "fairseq"
 cache_dir = (
     cache_dir or (Path.home() / ".cache" / library_name).as_posix()