Update app.py

app.py

@@ -1,43 +1,90 @@
 import gradio as gr
+import numpy as np
 from kittentts import KittenTTS
 
-# Initialize the KittenTTS model
-# This model is lightweight and runs on the CPU
-m = KittenTTS("KittenML/kitten-tts-nano-0.1")
+# 1. Initialize the KittenTTS model.
+# This will download the model from Hugging Face on the first run.
+print("Loading KittenTTS model...")
+try:
+    tts_model = KittenTTS("KittenML/kitten-tts-nano-0.1")
+    print("Model loaded successfully.")
+except Exception as e:
+    print(f"Error loading model: {e}")
+    # You might want to handle this more gracefully
+    exit()
 
-def text_to_speech(text):
-    """
-    Generates audio from the input text using the KittenTTS model.
 
-    Args:
-        text: The text to be converted to speech.
+# 2. Get the list of available voices directly from the model instance.
+AVAILABLE_VOICES = tts_model.available_voices
+DEFAULT_VOICE = "expr-voice-5-m" if "expr-voice-5-m" in AVAILABLE_VOICES else AVAILABLE_VOICES[0]
 
-    Returns:
-        A tuple containing the sample rate and the audio data as a NumPy array.
+# 3. Define the core function that Gradio will call.
+# This function now accepts 'voice' and 'speed' as arguments.
+def synthesize_speech(text, voice, speed):
+    """
+    Generates audio using the selected text, voice, and speed.
     """
-    # The KittenTTS model generates audio at a sample rate of 24000 Hz
-    sampling_rate = 24000
-    audio_numpy = m.generate(text)
-    return (sampling_rate, audio_numpy)
-
-# Define the Gradio interface
-iface = gr.Interface(
-    fn=text_to_speech,
-    inputs=gr.Textbox(
-        lines=3,
-        label="Text to Synthesize",
-        placeholder="Enter your text here..."
-    ),
-    outputs=gr.Audio(label="Synthesized Speech"),
-    title="KittenTTS: Text-to-Speech",
-    description="A simple Gradio app to demonstrate the capabilities of the KittenTTS model. KittenTTS is a lightweight, high-quality text-to-speech model that can run on a CPU. [2]",
-    examples=[
-        ["This high quality TTS model works without a GPU"],
-        ["Gradio is a great tool for creating machine learning demos."],
-        ["The quick brown fox jumps over the lazy dog."]
-    ],
-    allow_flagging="never"
-)
+    # Handle empty input gracefully
+    if not text.strip():
+        # Return a silent, empty audio clip
+        return (24000, np.zeros(0, dtype=np.int16))
+
+    # Call the model's generate method with all the parameters
+    audio_data = tts_model.generate(text, voice=voice, speed=speed)
+    
+    # Return the audio in the format Gradio expects: (sample_rate, numpy_array)
+    return (24000, audio_data)
+
+# 4. Create the Gradio UI with the new controls.
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown(
+        """
+        # 🐱 Enhanced KittenTTS UI
+        A user-friendly interface for the KittenTTS text-to-speech model.
+        Select a voice, adjust the speed, and type your text to generate audio.
+        """
+    )
+
+    with gr.Row():
+        with gr.Column(scale=3):
+            text_input = gr.Textbox(
+                lines=5,
+                label="Input Text",
+                placeholder="Type something here..."
+            )
+        
+        with gr.Column(scale=1):
+            voice_dropdown = gr.Dropdown(
+                choices=AVAILABLE_VOICES,
+                value=DEFAULT_VOICE,
+                label="Voice Selection"
+            )
+            speed_slider = gr.Slider(
+                minimum=0.5,
+                maximum=2.0,
+                step=0.1,
+                value=1.0,
+                label="Speech Speed"
+            )
+            generate_button = gr.Button("Generate Audio", variant="primary")
+            
+    audio_output = gr.Audio(label="Generated Speech", autoplay=True)
+
+    gr.Examples(
+        examples=[
+            ["This is an example of a female voice.", "expr-voice-5-f", 1.0],
+            ["This is an example of a male voice, speaking a bit faster.", "expr-voice-5-m", 1.2],
+            ["The speed can also be slowed down for clarity.", "expr-voice-4-f", 0.8],
+        ],
+        inputs=[text_input, voice_dropdown, speed_slider]
+    )
+
+    # Connect the UI components to the function
+    generate_button.click(
+        fn=synthesize_speech,
+        inputs=[text_input, voice_dropdown, speed_slider],
+        outputs=audio_output
+    )
 
 # Launch the Gradio app
-iface.launch()
+demo.launch()