Update app.py

app.py

@@ -1,31 +1,37 @@
 import torch
+import os
 import gradio as gr
 from bark import generate_audio, SAMPLE_RATE
 from scipy.io.wavfile import write as write_wav
-import numpy as np
-import os
+import shutil
 
-# Fix for PyTorch 2.6+ loading bug
-import numpy.core.multiarray
-import torch.serialization
+# Make sure history_prompts directory exists
+os.makedirs("history_prompts", exist_ok=True)
 
-torch.serialization._use_new_zipfile_serialization = False
+def synthesize(text_prompt, demo_voice=None):
+    history_prompt = None
 
-def synthesize(text_prompt, history_prompt=None):
-    audio_array = generate_audio(text_prompt, history_prompt=history_prompt)
-    output_path = "output.wav"
-    write_wav(output_path, SAMPLE_RATE, audio_array)
-    return output_path
+    # If user uploads a demo voice, save it and use as history prompt
+    if demo_voice is not None:
+        uploaded_path = "history_prompts/user_prompt.wav"
+        shutil.copyfile(demo_voice, uploaded_path)
+        history_prompt = uploaded_path
 
-demo = gr.Interface(
+    try:
+        audio_array = generate_audio(text_prompt, history_prompt=history_prompt)
+        output_path = "output.wav"
+        write_wav(output_path, SAMPLE_RATE, audio_array)
+        return output_path
+    except Exception as e:
+        return f"Error: {e}"
+
+gr.Interface(
     fn=synthesize,
     inputs=[
-        gr.Textbox(label="Enter Text to Speak"),
-        gr.Textbox(label="History Prompt Name (optional, e.g. v2/en_speaker_6)")
+        gr.Textbox(label="💬 Enter Text to Speak"),
+        gr.Audio(source="upload", type="filepath", label="📤 Upload Real Voice (optional)", optional=True)
     ],
-    outputs=gr.Audio(label="Generated Voice", type="filepath"),
-    title="CloneVoicetts - Real Voice Cloning (Bark)",
-    description="Upload a demo voice to train as history prompt (optional), then enter any text to generate speech in that style.",
-)
-
-demo.launch()
+    outputs=gr.Audio(type="filepath", label="🔊 Cloned Voice Output"),
+    title="🧬 CloneVoiceTTS - Real Voice Cloning (Bark)",
+    description="Upload a short real voice (.wav), then type any text. Bark will try to clone the uploaded voice tone."
+).launch()