Update app.py

app.py

@@ -1,33 +1,50 @@
 import gradio as gr
-from gtts import gTTS
+from TTS.api import TTS
 import tempfile
+import librosa
+import soundfile as sf
 
-def text_to_speech(text, voice_sample=None):
-    """
-    Generate Hindi speech from input text using gTTS.
-    Note: The voice_sample is accepted but not used, since voice cloning for Hindi is not supported.
-    """
-    # Generate Hindi speech using gTTS (lang='hi')
-    tts = gTTS(text=text, lang='hi')
+# Initialize Coqui TTS model for Hindi (IndicTTS)
+tts_model_name = "tts_models/multilingual/multi-dataset/xtts_v2"
+tts = TTS(tts_model_name)
+
+def text_to_speech_clone(text, voice_sample):
+    if voice_sample is None:
+        return "Please provide a voice sample audio.", None
     
-    # Save to a temporary MP3 file
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as fp:
-        tts.save(fp.name)
-        audio_file = fp.name
-    return audio_file
+    # Load voice sample
+    sample_wav, sample_rate = librosa.load(voice_sample, sr=22050)
+
+    # Save voice sample as wav temporarily
+    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp_sample:
+        sf.write(tmp_sample.name, sample_wav, sample_rate)
+        voice_sample_path = tmp_sample.name
+
+    # Generate cloned speech in Hindi
+    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp_output:
+        tts.tts_to_file(
+            text=text,
+            file_path=tmp_output.name,
+            speaker_wav=voice_sample_path,
+            language="hi"
+        )
+        output_path = tmp_output.name
+
+    return output_path
 
 iface = gr.Interface(
-    fn=text_to_speech,
+    fn=text_to_speech_clone,
     inputs=[
         gr.Textbox(lines=5, placeholder="हिंदी में टेक्स्ट दर्ज करें...", label="Text"),
-        gr.Audio(type="filepath", label="Voice Sample (ignored)")
+        gr.Audio(type="filepath", label="Voice Sample (Hindi speech)")
     ],
-    outputs=gr.Audio(type="filepath", label="Generated Speech"),
-    title="Hindi Text-to-Speech",
+    outputs=gr.Audio(type="filepath", label="Generated Cloned Speech"),
+    title="Hindi Text-to-Speech with Voice Cloning",
     description=(
-        "Generate Hindi speech from text using gTTS. Note: "
-        "Voice cloning is not available for Hindi; the voice sample input is ignored."
+        "Generate Hindi speech from text with voice cloning capability. "
+        "Provide a Hindi voice sample to clone its voice characteristics."
     )
 )
 
 iface.launch()
+