Update app.py

app.py

@@ -37,11 +37,12 @@ vocoder_model.load_state_dict(checkpoint['model_g'])
 vocoder_model.eval(inference=False)
 
 
-def synthesize(text):
+def synthesize(text, gst_1, gst_2, gst_3):
     sequence = np.array(text_to_sequence(text, ['english_cleaners']))[None, :]
     sequence = torch.from_numpy(sequence).to(device='cpu', dtype=torch.int64)
 
-    gst_head_scores = np.array([0.5, 0.15, 0.35])  # originally ([0.5, 0.15, 0.35])
+    # gst_head_scores = np.array([0.5, 0.15, 0.35])  # originally ([0.5, 0.15, 0.35])
+    gst_head_scores = np.array([gst_1, gst_2, gst_3])  # originally ([0.5, 0.15, 0.35])
     gst_scores = torch.from_numpy(gst_head_scores).float()
 
     mel_outputs, mel_outputs_postnet, _, alignments = model.inference(sequence, gst_scores)
@@ -54,6 +55,6 @@ def synthesize(text):
     return (22050, audio_numpy)
 
 
-iface = gr.Interface(fn=synthesize, inputs="text", outputs=[gr.Audio(label="Generated Speech", type="numpy"),])
+iface = gr.Interface(fn=synthesize, inputs=["text", gr.Slider(0.25, 0.55), gr.Slider(0.25, 0.55), gr.Slider(0.25, 0.55)], outputs=[gr.Audio(label="Generated Speech", type="numpy"),])
 iface.launch()