Update app.py

app.py

@@ -1,42 +1,56 @@
-from Crypto.Cipher import AES
-from Crypto.Protocol.KDF import PBKDF2
-import os
+import gradio as gr
+import torch
 import tempfile
-from dotenv import load_dotenv
-
-load_dotenv()  # Load all environment variables
-
-def unpad(data):
-    return data[:-data[-1]]
-
-def decrypt_and_run():
-    # Get password from Hugging Face Secrets environment variable
-    password = os.getenv("PASSWORD")
-    if not password:
-        raise ValueError("PASSWORD secret not found in environment variables")
-
-    password = password.encode()
-
-    with open("code.enc", "rb") as f:
-        encrypted = f.read()
-
-    salt = encrypted[:16]
-    iv = encrypted[16:32]
-    ciphertext = encrypted[32:]
-
-    key = PBKDF2(password, salt, dkLen=32, count=1000000)
-    cipher = AES.new(key, AES.MODE_CBC, iv)
-
-    plaintext = unpad(cipher.decrypt(ciphertext))
-
-    with tempfile.NamedTemporaryFile(suffix=".py", delete=False, mode='wb') as tmp:
-        tmp.write(plaintext)
-        tmp.flush()
-        print(f"[INFO] Running decrypted code from {tmp.name}")
-        os.system(f"python {tmp.name}")
+import soundfile as sf
+from tortoise.api import TextToSpeech
+from tortoise.utils.audio import load_audio
+
+# 1) Initialize the Tortoise TTS engine at startup
+tts = TextToSpeech()  # Downloads and caches models automatically
+
+# 2) Define a helper to generate speech from a reference clip + text
+def generate_speech(reference_audio_path, text):
+    """
+    reference_audio_path: filepath to a WAV sampled at 22 050 Hz
+    text: the string to synthesize
+    returns: path to a 24 kHz WAV file with your cloned voice
+    """
+    # ✅ FIXED: Provide sampling_rate as a required positional argument
+    ref_waveform = load_audio(reference_audio_path, 22050)
+
+    # Generate speech using 'fast' preset (alternatives: ultra_fast, standard, high_quality)
+    output_tensor = tts.tts_with_preset(
+        text,
+        voice_samples=[ref_waveform],
+        preset="fast"
+    )
+
+    # Save to temp WAV (float32, 24 kHz)
+    wav_np = output_tensor.squeeze().cpu().numpy()
+    tmp = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+    sf.write(tmp.name, wav_np, samplerate=24000)
+    return tmp.name
+
+# 3) Build the Gradio interface
+with gr.Blocks(title="Tortoise Voice Cloning TTS") as app:
+    gr.Markdown("## 🗣️ Voice Cloning with Tortoise TTS")
+    gr.Markdown(
+        "Upload a ~10 sec WAV clip (22 050 Hz), enter English text, "
+        "and hear it spoken back in **your** voice!"
+    )
+
+    with gr.Row():
+        voice_sample = gr.Audio(type="filepath", label="🎙️ Upload Reference Voice (22 050 Hz WAV)")
+        text_input   = gr.Textbox(label="💬 Text to Synthesize", placeholder="e.g., Hello, world!")
+
+    generate_btn = gr.Button("🔊 Generate Speech")
+    output_audio = gr.Audio(label="📢 Cloned Speech Output (24 kHz)", interactive=False)
+
+    generate_btn.click(
+        fn=generate_speech,
+        inputs=[voice_sample, text_input],
+        outputs=output_audio
+    )
 
 if __name__ == "__main__":
-    decrypt_and_run()
-
-# This script decrypts the encrypted code and runs it.
-# Ensure you have the PASSWORD secret set in your Hugging Face Secrets
+    app.launch()