Update app.py

app.py

@@ -1,10 +1,17 @@
 import gradio as gr
 import os
+from transformers import AutoTokenizer,VitsModel
 
 import google.generativeai as genai
-api_key =os.environ.get("id_gmkey")
+import torch
 
+api_key =os.environ.get("id_gmkey")
+token=os.environ.get("key_")
 genai.configure(api_key=api_key)
+tokenizer = AutoTokenizer.from_pretrained("asg2024/vits-ar-sa-huba",token=token)
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model=VitsModel.from_pretrained("asg2024/vits-ar-sa-huba",token=api_key).to(device)
+
 
 generation_config = {
   "temperature": 1,
@@ -79,6 +86,41 @@ def   get_answer_ai(text):
           AI=create_chat_session()
           response = AI.send_message(text,stream=True)
           return response
+
+def   modelspeech(text):
+     with torch.no_grad():
+          inputs = tokenizer(text, return_tensors="pt")#.cuda()
+
+          wav = model(input_ids=inputs["input_ids"].to(device)).waveform.cpu().numpy().reshape(-1)
+          # display(Audio(wav, rate=model.config.sampling_rate))
+          return  model.config.sampling_rate,wav#remove_noise_nr(wav)
+
+import re
+def clean_text(text):
+  # Remove symbols and extra spaces
+  cleaned_text = re.sub(r'[^\w\s]', '', text)  # Remove symbols
+  cleaned_text = re.sub(r'\s+', '  ', cleaned_text)  # Normalize spaces
+  return cleaned_text.strip()  # Remove leading/trailing spaces
+
+
+def text_to_speech(text):
+    
+    job = dash(text)
+    pad_text=''
+    k=0
+    for chunk in job:
+       
+       pad_text+=str(clean_text(chunk))
+       
+       if pad_text!='' and len(pad_text)>10:
+           out=pad_text
+           pad_text=''
+           k+=1
+        
+           yield modelspeech(out)
+    if k==0:
+         out=pad_text
+         yield modelspeech(pad_text)
 def dash(text):
     
     response=get_answer_ai(text)
@@ -87,5 +129,20 @@ def dash(text):
     # return textai
 
 
-demo = gr.Interface(fn=dash, inputs=["text"], outputs=['text'])
-demo.launch()
+# demo = gr.Interface(fn=dash, inputs=["text"], outputs=['text'])
+# demo.launch()
+
+with gr.Blocks() as demo:
+    with gr.Tab("AI Text  "):
+        gr.Markdown("# Text to Speech")
+        text_input = gr.Textbox(label="Enter Text")
+        text_out = gr.Textbox()
+        text_input.submit(dash, text_input, text_out)
+    with gr.Tab("AI Speech"):
+        gr.Markdown("# Text to Speech")
+        text_input2 = gr.Textbox(label="Enter Text")
+        audio_output = gr.Audio(streaming=True)
+        text_input2.submit(text_to_speech, text_input2, audio_output)
+         
+
+demo.launch(show_error=True)