Update xtts.py

xtts.py

@@ -54,21 +54,11 @@ if tts is None:
     model.get_conditioning_latents=get_conditioning_latents
     logging.info("model is ready")
 
-@app.route("/convert")
-def predict():
+def predict(text, sample=None, language="zh"):
     global tts 
     global model
-
-    text = request.args.get('text')
-    sample = request.args.get('sample')
-    language = request.args.get('language')
-
-    if text is None:
-        return jsonify({'error': 'text is missing'}), 400
-    
-    text= re.sub("([^\x00-\x7F]|\w)(\.|\。|\?)",r"\1 \2\2",text)
-    
     try:
+        text= re.sub("([^\x00-\x7F]|\w)(\.|\。|\?)",r"\1 \2\2",text)
         wav = tts.tts(
             text, 
             language=language if language is not None else "zh",
@@ -92,6 +82,17 @@ def predict():
         return str(e)
 
 @app.route("/")
+def convert():
+    text = request.args.get('text')
+    if text is None:
+        return jsonify({'error': 'text is missing'}), 400
+        
+    sample = request.args.get('sample')
+    language = request.args.get('language')
+    
+    return predict(text, sample, language)
+
+@app.route("/play")
 def play():
     url=predict()
     return f'''
@@ -155,4 +156,7 @@ def trim_sample_audio(speaker_wav):
         traceback.print_exc()
         return speaker_wav
 
-logging.info("xtts is ready")
+logging.info("xtts is ready")
+
+import gradio as gr
+gr.Interface(predict, inputs=["text", "text"], outputs=gr.Audio()).launch()