Update app.py

app.py

@@ -1,57 +1,26 @@
 import gradio as gr
-import librosa
-import numpy as np
+from gtts import gTTS
+import tempfile
 
-# Try to import a helper function to load the TTS model.
-try:
-    from parler_tts import load_tts
-except ImportError:
-    raise ImportError(
-        "The function 'load_tts' could not be imported from parler_tts. "
-        "Please check the documentation or the installed package structure for the correct API."
-    )
-
-# Initialize the TTS model for Hindi with voice cloning enabled.
-# (Parameters may vary depending on the actual API.)
-tts_model = load_tts(language="hi", voice_cloning=True)
-
-def extract_speaker_embedding(voice_sample_path):
-    """
-    Extract a speaker embedding from an uploaded Hindi voice sample.
-    This function loads the audio file, resamples to 16 kHz, and extracts the speaker embedding.
-    """
-    wav, sr = librosa.load(voice_sample_path, sr=16000)
-    # Assuming the tts_model provides a method for embedding extraction.
-    speaker_embedding = tts_model.extract_embedding(wav)
-    return speaker_embedding
-
-def synthesize_voice_with_cloning(voice_sample_path, hindi_text):
-    """
-    Synthesize Hindi speech from text, cloning the voice characteristics from the uploaded sample.
-    """
-    # Extract the speaker embedding.
-    speaker_embedding = extract_speaker_embedding(voice_sample_path)
-    
-    # Synthesize speech using the provided text and speaker embedding.
-    audio_waveform = tts_model.synthesize(text=hindi_text, speaker_embedding=speaker_embedding)
-    
-    # Convert output to a numpy array if necessary.
-    if not isinstance(audio_waveform, np.ndarray):
-        audio_waveform = np.array(audio_waveform)
+def text_to_speech(text):
+    # Generate speech using gTTS with Hindi language ('hi')
+    tts = gTTS(text=text, lang='hi')
     
-    return audio_waveform
+    # Save the audio to a temporary file
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as fp:
+        tts.save(fp.name)
+        audio_file = fp.name
+        
+    return audio_file
 
-# Create a Gradio interface.
+# Create the Gradio interface
 iface = gr.Interface(
-    fn=synthesize_voice_with_cloning,
-    inputs=[
-        gr.Audio(source="upload", type="filepath", label="Upload Hindi Voice Sample"),
-        gr.Textbox(lines=3, placeholder="Enter Hindi text here...", label="Hindi Text")
-    ],
-    outputs=gr.Audio(label="Generated Speech"),
-    title="Hindi TTS with Voice Cloning",
-    description="Upload a Hindi voice sample and enter Hindi text to generate cloned speech."
+    fn=text_to_speech,
+    inputs=gr.Textbox(lines=5, placeholder="हिंदी में टेक्स्ट दर्ज करें...", label="Enter Hindi Text"),
+    outputs=gr.Audio(type="file", label="Generated Speech"),
+    title="Hindi Text-to-Speech",
+    description="Convert Hindi text into speech using gTTS."
 )
 
-if __name__ == "__main__":
-    iface.launch()
+# Launch the app
+iface.launch()