Update app.py

app.py

@@ -1,24 +1,54 @@
 import gradio as gr
 from TTS.api import TTS
 import uuid
+import os
+from pydub import AudioSegment
 
-tts = TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", progress_bar=False, gpu=False)
+# Load XTTS v2 model (slow on CPU, suitable for ~10s voiceovers)
+tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", progress_bar=False, gpu=False)
 
-def generate_voice(script):
-    output_path = f"output_{uuid.uuid4().hex}.wav"
-    tts.tts_to_file(text=script, file_path=output_path)
-    return output_path, output_path
+# Define emotional styles mapping (XTTS uses style_wav or speaker embedding)
+EMOTIONS = {
+    "Neutral": None,
+    "Sad": "sad",
+    "Happy": "happy",
+    "Angry": "angry",
+    "Excited": "excited"
+}
 
+# Generate speech function
+def generate_voice(script, emotion):
+    # Output path as WAV first
+    wav_output_path = f"output_{uuid.uuid4().hex}.wav"
+    
+    # XTTS v2 supports emotion as "style" param
+    tts.tts_to_file(
+        text=script,
+        file_path=wav_output_path,
+        speaker_wav=None,  # You can also add your own voice sample here
+        language="en",
+        emotion=EMOTIONS[emotion]  # Optional style tag
+    )
+
+    # Convert to MP3 using pydub
+    mp3_output_path = wav_output_path.replace(".wav", ".mp3")
+    audio = AudioSegment.from_wav(wav_output_path)
+    audio.export(mp3_output_path, format="mp3")
+
+    return mp3_output_path, mp3_output_path
+
+# Gradio UI
 with gr.Blocks() as demo:
-    gr.Markdown("## 🎙️ AI Voiceover Generator")
+    gr.Markdown("## 🎙️ AI Voiceover Generator with Emotions")
 
-    script_input = gr.Textbox(label="Enter Script", placeholder="Type here...", lines=4, max_lines=4, max_length=200)
+    script_input = gr.Textbox(label="🎤 Script", placeholder="Enter your voiceover script...", lines=4, max_lines=4)
+    emotion_input = gr.Dropdown(label="🎭 Emotion", choices=list(EMOTIONS.keys()), value="Neutral")
     generate_btn = gr.Button("Generate Voice")
 
-    audio_output = gr.Audio(label="Preview", type="filepath")
-    download_output = gr.File(label="Download")
+    audio_output = gr.Audio(label="🎧 Listen", type="filepath")
+    download_output = gr.File(label="⬇️ Download MP3")
 
-    generate_btn.click(fn=generate_voice, inputs=script_input, outputs=[audio_output, download_output])
+    generate_btn.click(fn=generate_voice, inputs=[script_input, emotion_input], outputs=[audio_output, download_output])
 
 if __name__ == "__main__":
     demo.launch()