Update README.md

README.md

@@ -1,12 +1,39 @@
 ## Example to download fastspeech2 from fairseq
 
-Weights are downloaded from:
-
-We still need to git clone this repo first before being able to download it.
-Having `cd`'ed into the repo we can do the following:
+The following should work with fairseq's most up-to-date version in a google colab:
 
 ```python
 from fairseq.checkpoint_utils import load_model_ensemble_and_task_from_hf
 
-model = load_model_ensemble_and_task_from_hf("patrickvonplaten/fairseq-fastspeech2")
-```
+model_ensemble, cfg, task = load_model_ensemble_and_task_from_hf_hub(
+    "facebook/fastspeech2-test", arg_overrides={"vocoder": "griffin_lim", "fp16": False}
+)
+
+def tokenize(text):
+  import g2p_en
+  tokenized = g2p_en.G2p()(text)
+  tokenized = [{",": "sp", ";": "sp"}.get(p, p) for p in tokenized]
+  return " ".join(p for p in tokenized if p.isalnum())
+  
+text = "This is a cool demo for speech synthesis, don't you think so?"
+
+tokenized = tokenize(text)
+sample = {
+    "net_input": {
+        "src_tokens": task.src_dict.encode_line(tokenized).view(1, -1),
+        "src_lengths": torch.Tensor([len(tokenized.split())]).long(),
+        "prev_output_tokens": None
+        },
+    "target_lengths": None,
+    "speaker": None,
+}
+generator = task.build_generator(model_ensemble, cfg)
+generation = generator.generate(model_ensemble[0], sample)
+waveform = generation[0]["waveform"]
+
+import IPython.display as ipd
+
+ipd.Audio(waveform, rate=task.sr)
+```
+
+See: https://colab.research.google.com/drive/1gvq4Y1urrg9QrQ9031sZIP93LKspIh_X?usp=sharing