Update app.py

app.py

@@ -38,7 +38,8 @@ def predict(text, preset):
     input_ids = inputs["input_ids"]
     input_ids = input_ids[..., :model.config.max_text_positions]
 
-    speaker_embedding = np.load('spkemb/cmu_us_awb_arctic-wav-arctic_a0002.npy', allow_pickle=True)
+    # cmu_us_awb_arctic-wav-arctic_a0002.npy
+    speaker_embedding = np.load('spkemb/cmu_us_bdl_arctic-wav-arctic_a0009.npy', allow_pickle=True)
     speaker_embedding = torch.tensor(speaker_embedding).unsqueeze(0)
     speech = model.generate_speech(input_ids, speaker_embedding, vocoder=vocoder)
 
@@ -46,11 +47,10 @@ def predict(text, preset):
     return (16000, speech)
 
 
-title = "SpeechT5: Speech Synthesis"
+title = "Create 423: News to Speech"
 
 description = """
-The <b>SpeechT5</b> model is pre-trained on text as well as speech inputs, with targets that are also a mix of text and speech.
-By pre-training on text and speech at the same time, it learns unified representations for both, resulting in improved modeling capabilities.
+Create 423: News to Speech
 """
 
 article = """