Text to speech_1

app.py

@@ -1,7 +1,46 @@
 import gradio as gr
+import torch
+from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
+import soundfile as sf
+import tempfile
+import os
 
-def greet(name):
-    return "Hello " + name + "!!"
+# Load processor and model
+processor = AutoProcessor.from_pretrained("ai4bharat/indic-parler-tts")
+model = AutoModelForSpeechSeq2Seq.from_pretrained("ai4bharat/indic-parler-tts")
+model.eval()
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+LANGUAGE_OPTIONS = {
+    "Hindi": "hi", "Tamil": "ta", "Telugu": "te", "Malayalam": "ml", "Kannada": "kn",
+    "Bengali": "bn", "Marathi": "mr", "Gujarati": "gu", "Punjabi": "pa",
+    "Odia": "or", "Assamese": "as", "Urdu": "ur", "English (Indian)": "en"
+}
+
+def tts_generate(text, language_name):
+    lang = LANGUAGE_OPTIONS[language_name]
+    
+    inputs = processor(text=[text], return_tensors="pt", lang=lang)
+    with torch.no_grad():
+        output = model.generate(**inputs)
+    
+    audio_arr = processor.decode(output[0], skip_special_tokens=True)
+    
+    # Save audio as temporary .wav file
+    temp_path = tempfile.mktemp(suffix=".wav")
+    sf.write(temp_path, audio_arr, 16000)
+    
+    return temp_path
+
+# Gradio Interface
+interface = gr.Interface(
+    fn=tts_generate,
+    inputs=[
+        gr.Textbox(label="Enter Text"),
+        gr.Dropdown(choices=list(LANGUAGE_OPTIONS.keys()), label="Select Language")
+    ],
+    outputs=gr.Audio(label="Generated Audio", type="filepath"),
+    title="Indic Parler TTS - AI4Bharat",
+    description="Enter text and choose a language to generate and download speech audio."
+)
+
+interface.launch()