Update app.py

app.py

@@ -190,7 +190,13 @@ def   modelspeech(text):
           wav = model_vits(input_ids=inputs["input_ids"]).waveform.cpu().numpy().reshape(-1)
           # display(Audio(wav, rate=model.config.sampling_rate))
           return  model_vits.config.sampling_rate,wav#remove_noise_nr(wav)
+def   modelspeechstr(text):
+     with torch.no_grad():
+          inputs = tokenizer(text, return_tensors="pt")#.cuda()
 
+          wav = model_vits(input_ids=inputs["input_ids"]).waveform.cpu().numpy().reshape(-1)
+          # display(Audio(wav, rate=model.config.sampling_rate))
+          return  np.array2string(wav)
 import re
 def clean_text(text):
   # Remove symbols and extra spaces
@@ -217,8 +223,30 @@ def text_to_speech(text,session_ai):
            #      yield stream_wav
     if pad_text!='':
         yield modelspeech(pad_text),session_ai
+
        # for   stream_wav in generate_audio(pad_text):
        #          yield stream_wav
+
+
+
+def text_to_speechstr(text,session_ai):
+    
+    response = dash(text,session_ai,False)
+    pad_text=''
+    k=0
+    for chunk in response:
+       chunk,session_ai=chunk
+       pad_text+=str(clean_text(chunk))
+       
+       if pad_text!='' and len(pad_text)>10:
+           out=pad_text
+           pad_text=''
+           k+=1
+           yield modelspeechstr(out),session_ai
+           # for   stream_wav in generate_audio(out):
+           #      yield stream_wav
+    if pad_text!='':
+        yield modelspeechstr(pad_text),session_ai
 def dash(text,session_ai,is_state=True):
     
     response,session_ai=get_answer_ai(text,session_ai)
@@ -251,6 +279,11 @@ with gr.Blocks() as demo:
         text_input2 = gr.Textbox(label="Enter Text")
         audio_output = gr.Audio(streaming=True,autoplay=True)
         text_input2.submit(text_to_speech, [text_input2,session_ai], [audio_output,session_ai])
+    with gr.Tab("AI Speechstr"):
+        gr.Markdown("# Text to Speech")
+        text_input3 = gr.Textbox(label="Enter Text")
+        text_input4 = gr.Textbox(label="out Text")
+        text_input3.submit(text_to_speechstr, [text_input3,session_ai], [text_input4,session_ai])
          
 
 demo.launch(show_error=True)