Update app.py

app.py

@@ -1,23 +1,61 @@
-import spaces
-import gradio as gr
+import streamlit as st
 import torch
 from TTS.api import TTS
 import os
+import tempfile
+
 os.environ["COQUI_TOS_AGREED"] = "1"
 
-device = "cuda"
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# Initialize TTS model
+@st.cache_resource
+def load_tts_model():
+    return TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
+
+tts = load_tts_model()
+
+def clone(text, audio_file, language, speaking_rate, pitch, volume, 
+          emotion, sample_rate, temperature, seed):
+    if seed is not None:
+        torch.manual_seed(seed)
+    
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
+        temp_audio_path = temp_audio.name
+    
+    tts.tts_to_file(
+        text=text, 
+        speaker_wav=audio_file,
+        language=language,
+        file_path=temp_audio_path
+    )
+    
+    return temp_audio_path
+
+st.title('Advanced Voice Clone')
+st.write('Customize your voice cloning experience with various parameters.')
 
-tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
+text = st.text_area('Text')
+audio_file = st.file_uploader('Voice reference audio file', type=['wav', 'mp3'])
+language = st.selectbox('Language', ["en", "es", "fr", "de", "it"], index=0)
+speaking_rate = st.slider('Speaking Rate', 0.5, 2.0, 1.0)
+pitch = st.slider('Pitch Adjustment', -10, 10, 0)
+volume = st.slider('Volume', 0.1, 2.0, 1.0)
+emotion = st.selectbox('Emotion', ["neutral", "happy", "sad", "angry"], index=0)
+sample_rate = st.selectbox('Sample Rate', [22050, 24000, 44100, 48000], index=1)
+temperature = st.slider('Temperature', 0.1, 1.0, 0.8)
+seed = st.number_input('Seed (optional)', value=None)
 
-def clone(text, audio):
-    tts.tts_to_file(text=text, speaker_wav=audio, language="en", file_path="./output.wav")
-    return "./output.wav"
+if st.button('Generate'):
+    if text and audio_file:
+        with st.spinner('Generating audio...'):
+            output_path = clone(text, audio_file, language, speaking_rate, pitch, volume, 
+                                emotion, sample_rate, temperature, seed)
+            st.audio(output_path)
+    else:
+        st.warning('Please provide both text and a voice reference audio file.')
 
-iface = gr.Interface(fn=clone, 
-                     inputs=[gr.Textbox(label='Text'),gr.Audio(type='filepath', label='Voice reference audio file')], 
-                     outputs=gr.Audio(type='filepath'),
-                     title='Voice Clone',
-                     description="""
-                     """,
-                     theme = gr.themes.Base(primary_hue="teal",secondary_hue="teal",neutral_hue="slate"))
-iface.launch()
+# Clean up temporary files
+for file in os.listdir():
+    if file.endswith('.wav') and file.startswith('tmp'):
+        os.remove(file)