Create app.py

app.py

@@ -0,0 +1,225 @@
+import gradio as gr
+import os
+import re
+import uuid
+import scipy.io.wavfile
+import torch
+from pocket_tts import TTSModel
+
+print("Loading TTS Model...")
+try:
+    tts_model = TTSModel.load_model()
+    print("Model loaded successfully.")
+except Exception as e:
+    print(f"Error loading model: {e}")
+
+def get_tts_file_name(text, language="en"):
+    temp_audio_dir = "./ai_tts_voice/"
+    os.makedirs(temp_audio_dir, exist_ok=True)
+
+    clean = re.sub(r'[^a-zA-Z\s]', '', text or "")
+    clean = clean.lower().strip().replace(" ", "_")[:20] or "audio"
+
+    uid = uuid.uuid4().hex[:8].upper()
+    language = language.lower().strip()
+
+    return os.path.join(
+        temp_audio_dir,
+        f"{clean}_{language}_{uid}.wav"
+    )
+
+DEFAULT_VOICES = [
+    "alba", "marius", "javert", "jean", 
+    "fantine", "cosette", "eponine", "azelma"
+]
+
+def generate_speech(text, mode, preset_voice, clone_audio_path):
+    if not text:
+        raise gr.Error("Please enter text to generate speech.")
+
+    state = None
+    
+    if mode == "Default Voices":
+        print(f"Using preset voice: {preset_voice}")
+        state = tts_model.get_state_for_audio_prompt(preset_voice)
+        
+    else:
+        if not clone_audio_path:
+            raise gr.Error("Please upload a reference audio file for cloning.")
+        
+        print(f"Cloning voice from: {clone_audio_path}")
+        try:
+            state = tts_model.get_state_for_audio_prompt(clone_audio_path)
+        except Exception as e:
+            error_msg = f"Error loading reference audio: {str(e)}. Please upload a valid WAV file."
+            print(error_msg)
+            raise gr.Error(error_msg)
+
+    try:
+        audio_tensor = tts_model.generate_audio(state, text)
+        
+        output_filename = get_tts_file_name(text)
+        scipy.io.wavfile.write(output_filename, tts_model.sample_rate, audio_tensor.numpy())
+        
+        return output_filename
+    except Exception as e:
+        raise gr.Error(f"Generation failed: {str(e)}")
+
+def toggle_inputs(mode):
+    if mode == "Default Voices":
+        return gr.update(visible=True), gr.update(visible=False)
+    else:
+        return gr.update(visible=False), gr.update(visible=True)
+
+
+CUSTOM_CSS = """
+.gradio-container { 
+    font-family: 'SF Pro Display', -apple-system, BlinkMacSystemFont, sans-serif; 
+}
+.header-container {
+    text-align: center;
+    margin-bottom: 20px;
+}
+.logo-img {
+    margin: 0 auto;
+    display: block;
+    max-width: 100%;
+    transition: transform 0.2s;
+}
+.logo-img:hover {
+    transform: scale(1.02);
+    opacity: 0.9;
+}
+.links-container a {
+    text-decoration: none; 
+    color: #4a90e2; 
+    font-weight: 500;
+}
+.links-container a:hover {
+    text-decoration: underline;
+}
+"""
+
+HEADER_HTML = """
+<div class="header-container" style="text-align:center;">
+
+    <a href="https://kyutai.org/tts" target="_blank" title="Visit Kyutai TTS">
+        <img src="https://raw.githubusercontent.com/kyutai-labs/pocket-tts/refs/heads/main/docs/logo.png"
+             class="logo-img" width="200">
+    </a>
+
+    <div class="links-container"
+         style="
+            margin-top: 18px;
+            display: flex;
+            justify-content: center;
+            align-items: center;
+            gap: 14px;
+            flex-wrap: wrap;
+         ">
+
+        <a href="https://github.com/kyutai-labs/pocket-tts"
+           target="_blank"
+           style="text-decoration:none;">
+            🐱 GitHub Repository
+        </a>
+
+        <span style="color: gray;">|</span>
+
+        <a href="https://huggingface.co/kyutai/pocket-tts"
+           target="_blank"
+           style="text-decoration:none;">
+            🤗 Hugging Face Model Card
+        </a>
+
+        <span style="color: gray;">|</span>
+
+        <a href="https://colab.research.google.com/github/NeuralFalconYT/Voice-Clone/blob/main/Pocket_TTS_Colab.ipynb"
+           target="_blank"
+           style="
+                display: inline-flex;
+                align-items: center;
+           ">
+            <img src="https://colab.research.google.com/assets/colab-badge.svg"
+                 alt="Open in Colab"
+                 height="26">
+        </a>
+
+    </div>
+
+    <p style="font-size: 0.8em; color: gray; margin-top: 10px;">
+        <i>Note: This is not an official demo from Kyutai Labs</i>
+    </p>
+
+</div>
+"""
+
+
+
+with gr.Blocks(theme='JohnSmith9982/small_and_pretty', css=CUSTOM_CSS) as demo:
+    gr.HTML(HEADER_HTML)
+    
+    with gr.Row():
+        with gr.Column():
+            text_input = gr.Textbox(
+                label="Text Input", 
+                placeholder="Hi, how are you?", 
+                lines=3,
+                value="Hi, how are you?"
+            )
+            
+            mode_radio = gr.Radio(
+                choices=["Default Voices", "Voice Clone"],
+                value="Default Voices",
+                label="TTS Mode"
+            )
+            
+            with gr.Group():
+                dropdown_input = gr.Dropdown(
+                    choices=DEFAULT_VOICES,
+                    value="alba",
+                    label="Select Voice",
+                    visible=True
+                )
+                
+                audio_upload = gr.Audio(
+                    label="Upload Reference Audio (WAV recommended)",
+                    type="filepath",
+                    visible=False
+                )
+            
+            generate_btn = gr.Button("Generate Audio", variant="primary")
+            
+            example_audio_url = "https://huggingface.co/kyutai/tts-voices/resolve/main/alba-mackenna/casual.wav"
+            
+            gr.Examples(
+                examples=[
+                    ["Hello, I am Fantine. Nice to meet you.", "Default Voices", "fantine", None],
+                    ["I am Cosette, and the weather is lovely.", "Default Voices", "cosette", None],
+                    ["Hey there, Eponine here.", "Default Voices", "eponine", None],
+                    ["Greetings from Azelma.", "Default Voices", "azelma", None],
+                    ["This is a voice cloning test using the uploaded reference audio.", "Voice Clone", None, example_audio_url],
+                ],
+                inputs=[text_input, mode_radio, dropdown_input, audio_upload],
+                label="Click on an Example to Try"
+            )
+
+        with gr.Column():
+            output_audio = gr.Audio(label="Generated Speech", type="filepath")
+
+
+    mode_radio.change(
+        fn=toggle_inputs,
+        inputs=[mode_radio],
+        outputs=[dropdown_input, audio_upload]
+    )
+    
+    generate_btn.click(
+        fn=generate_speech,
+        inputs=[text_input, mode_radio, dropdown_input, audio_upload],
+        outputs=[output_audio]
+    )
+
+if __name__ == "__main__":
+    demo.queue().launch(share=False, debug=False)
+