Update app.py

app.py

@@ -38,11 +38,17 @@ def merge_audio_video(entities_num, resize_img_list, text_input):
     
     return mergedclip
 
-fastspeech = gr.Interface.load("huggingface/facebook/fastspeech2-en-ljspeech", api_key=HF_TOKEN)
-
 def text2speech(text):
-    speech = fastspeech(text)
-    return speech
+    # Generate speech from text using FastSpeech2
+    speech_output = fastspeech(text)
+    # Save the output as a .flac file (assuming the output is in numpy format)
+    with open("speech_output.flac", "wb") as f:
+        f.write(speech_output["audio"])
+    return "speech_output.flac"
+    
+# Load FastSpeech2 model from Hugging Face directly
+fastspeech = pipeline("text-to-speech", model="facebook/fastspeech2-en-ljspeech", use_auth_token=HF_TOKEN)
+
 
 def engine(text_input):
     ner = gr.Interface.load("huggingface/flair/ner-english-ontonotes-large", api_key=HF_TOKEN)
@@ -63,6 +69,11 @@ def engine(text_input):
     
     return 'mergedvideo.mp4'
 
+
+
+
+
+
 app = gr.Interface(
     fn=engine, 
     inputs=gr.Textbox(lines=5, label="Input Text"),