Update app.py

app.py

@@ -1,32 +1,84 @@
-import spaces
-import gradio as gr
+
 import torch
 from TTS.api import TTS
 import os
+import docx2txt
+import gradio as gr
+import json
+
+folder = '.'
+
+
 os.environ["COQUI_TOS_AGREED"] = "1"
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
 
-@spaces.GPU(enable_queue=True)
-def clone(text, audio, language):
-    tts.tts_to_file(text=text, speaker_wav=audio, language=language, file_path="./output.wav")
-    return "./output.wav"
-
-iface = gr.Interface(fn=clone, 
-                     inputs=[gr.Textbox(label='Text'),gr.Audio(type='filepath', label='Voice reference audio file'),gr.Textbox(label='Langauge')], 
-                     outputs=gr.Audio(type='filepath'),
-                     title='Voice Clone',
-                     description="""
-                     by [Tony Assi](https://www.tonyassi.com/)
-                     
-                     Please ❤️ this Space. I build custom AI apps for companies. <a href="mailto: tony.assi.media@gmail.com">Email me</a> for business inquiries.
-                     """,
-                     theme = gr.themes.Base(primary_hue="teal",secondary_hue="teal",neutral_hue="slate"),
-                     examples=[["Hey! It's me Dorthy, from the Wizard of Oz. Type in whatever you'd like me to say.","./audio/Wizard-of-Oz-Dorthy.wav"],
-                               ["It's me Vito Corleone, from the Godfather. Type in whatever you'd like me to say.","./audio/Godfather.wav"],
-                               ["Hey, it's me Paris Hilton. Type in whatever you'd like me to say.","./audio/Paris-Hilton.mp3"],
-                               ["Hey, it's me Megan Fox from Transformers. Type in whatever you'd like me to say.","./audio/Megan-Fox.mp3"],
-                               ["Hey there, it's me Jeff Goldblum. Type in whatever you'd like me to say.","./audio/Jeff-Goldblum.mp3"],])
-iface.launch()
+folder = '.'
+
+with open(f"{folder}/config.json") as f:
+    config_settings = json.load(f)
+
+def generate_voiceover(text, voice, n, name):
+    if not os.path.exists(f"{folder}/Result/{voice}"):
+        os.makedirs(f"{folder}/Result/{voice}")
+    file_path = f"{folder}/Result/{voice}/{name}_{n}.mp3"
+    tts.tts_to_file(text=text,  speaker_wav=f'{folder}/voices/{voice}.mp3', language=config_settings["voices"][f'{voice}.mp3'], file_path= file_path)
+    # speed up for 1.2
+    os.system("play " +file_path+" tempo {}".format(config_settings["speed"]))
+
+    return file_path
+
+
+
+def process_file(file, voice):
+    if file is None:
+        return "Please upload a file.", None
+
+    file_type = file.name.split(".")[-1]
+
+    if file_type == "txt":
+        with open(file) as file_:
+          text = file_.read()
+    elif file_type == "docx":
+        text  = docx2txt.process()
+    else:
+        return "Unsupported file type.", None
+
+    if not text.strip():
+        return "File is empty. Please upload a file with content.", None
+
+    paragraphs = text.split("\n\n")
+    print(paragraphs)
+    audio_outputs = []
+    n=1
+    for paragraph in paragraphs:
+        if paragraph.strip():
+            audio_placeholder = generate_voiceover(paragraph, voice, n, file.name.split("/")[-1].split(".")[0])
+            audio_outputs.append(audio_placeholder)
+            n+=1
+
+    return "\n\n".join(audio_outputs)
+
+# all fileanmes in voices folder
+voices = [f.name.split(".")[0] for f in os.scandir("voices") if f.is_file()]
+
+
+def main():
+    gr.Interface(
+        fn=process_file,
+        inputs=[
+            gr.File(file_types=['.txt', '.docx'], label="Upload a file"),
+
+            gr.Dropdown(voices, label="Select a language")
+        ],
+        outputs=[
+            gr.Textbox(label="Voiceover Text")
+        ],
+        title="Text to Speech App",
+        description="Upload a .txt or .docx file, select a language, and generate voiceovers for the content."
+    ).launch(debug=True)
+
+if __name__ == "__main__":
+    main()