xVASynth synthesize return JSON

app.py

@@ -200,11 +200,16 @@ def predict(
 		response.raise_for_status()  # If the response contains an HTTP error status code, raise an exception
 	except requests.exceptions.RequestException as err:
 		print('Failed to synthesize!')
+		print('server.log contents:')
+		with open('resources/app/server.log', 'r') as f:
+			print(f.read())
+		return ['', err]
+
 	print('server.log contents:')
 	with open('resources/app/server.log', 'r') as f:
 		print(f.read())
 
-	return save_path
+	return [save_path, response]
 
 input_textbox = gr.Textbox(
 	label="Input Text",
@@ -259,7 +264,10 @@ gradio_app = gr.Interface(
 		sad_slider,
 		surprise_slider
 	],
-	outputs=gr.Audio(label="22kHz audio output", type="filepath"),
+	outputs=[
+		gr.Audio(label="22kHz audio output", type="filepath"),
+		"text"
+	],
 	title="xVASynth (WIP)",
 	clear_btn=gr.Button(visible=False)
 	# examples=[

resources/app/python/xvapitch/model.py

@@ -753,17 +753,19 @@ class xVAPitch(object):
             em_sad_pred.squeeze().cpu().detach().numpy() if em_sad_pred is not None else [],
             em_surprise_pred.squeeze().cpu().detach().numpy() if em_surprise_pred is not None else [],
         ]
-        editor_values_text = ",".join([str(v) for v in pitch]) + "\n" + \
-                             ",".join([str(v) for v in durations]) + "\n" + \
-                             ",".join([str(v) for v in energy]) + "\n" + \
-                             ",".join([str(v) for v in em_angry]) + "\n" + \
-                             ",".join([str(v) for v in em_happy]) + "\n" + \
-                             ",".join([str(v) for v in em_sad]) + "\n" + \
-                             ",".join([str(v) for v in em_surprise]) + "\n" + \
-                             json.dumps(editorStyles)
-
-        del pitch_pred, dur_pred, energy_pred, em_angry, em_happy, em_sad, em_surprise, text, sequence
-        return editor_values_text +"\n"+all_cleaned_text +"\n"+ f'{start_index}\n{end_index}'
+
+        del pitch_pred, dur_pred, energy_pred, text, sequence
+        return {
+            "pitch": pitch,
+            "durations": durations,
+            "energy": energy,
+            "em_angry": em_angry,
+            "em_happy": em_happy,
+            "em_sad": em_sad,
+            "em_surprise": em_surprise,
+            "editorStyles": json.dumps(editorStyles),
+            "arpabet": all_cleaned_text
+        }
 
     def set_device (self, device):
         self.device = device

resources/app/server.py

@@ -131,7 +131,7 @@ if __name__ == '__main__':
     class Handler(BaseHTTPRequestHandler):
         def _set_response(self):
             self.send_response(200)
-            self.send_header("Content-Type", "text/html")
+            self.send_header("Content-Type", "application/json")
             self.end_headers()
 
         def do_GET(self):