Create app.py

app.py

@@ -0,0 +1,66 @@
+# app.py
+import gradio as gr
+import torch
+import torchaudio
+from TTS.api import TTS
+import uuid
+import os
+
+# Load the XTTS model from Coqui
+model_name = "tts_models/multilingual/multi-dataset/xtts_v2"
+tts = TTS(model_name)
+
+# Emotions supported by XTTS
+EMOTION_MAP = {
+    "Neutral": "neutral",
+    "Sad": "sad",
+    "Happy": "happy",
+    "Angry": "angry",
+    "Excited": "excited"
+}
+
+# Output directory
+os.makedirs("outputs", exist_ok=True)
+
+def generate_voiceover(text, emotion):
+    emotion_label = EMOTION_MAP.get(emotion, "neutral")
+    
+    # Generate a temporary filename
+    output_path = f"outputs/{uuid.uuid4().hex}.wav"
+
+    # XTTS only supports cloning voice with reference. Use default speaker.
+    tts.tts_to_file(
+        text=text,
+        file_path=output_path,
+        speaker_wav=None,
+        language="en",
+        emotion=emotion_label
+    )
+
+    # Convert to MP3
+    mp3_path = output_path.replace(".wav", ".mp3")
+    waveform, sample_rate = torchaudio.load(output_path)
+    torchaudio.save(mp3_path, waveform, sample_rate, format="mp3")
+
+    return mp3_path
+
+# Gradio UI
+def app(text, emotion):
+    mp3_path = generate_voiceover(text, emotion)
+    return mp3_path, mp3_path
+
+iface = gr.Interface(
+    fn=app,
+    inputs=[
+        gr.Textbox(label="Enter your script here", lines=5, placeholder="Type something..."),
+        gr.Dropdown(label="Choose Emotion", choices=list(EMOTION_MAP.keys()), value="Neutral")
+    ],
+    outputs=[
+        gr.Audio(label="Generated Voiceover"),
+        gr.File(label="Download MP3")
+    ],
+    title="🎙️ AI Voiceover Generator with Emotion Control",
+    description="Convert your script into a voiceover with emotion using XTTS (free & open-source)."
+)
+
+iface.launch()