update

app.py

@@ -97,7 +97,7 @@ def whisper_stt(audio):
 
 
 # LLM - Bloom Response  
-def lang_model_response(prompt, language): 
+def lang_model_response(prompt, prompt_en, language): 
   print(f"Inside lang_model_response - Prompt is :{prompt}")
   p_en = """Question: How are you doing today?
   Answer: I am doing good, thanks.
@@ -109,18 +109,25 @@ def lang_model_response(prompt, language):
   Réponse: Je vais bien, merci.
   Question: """
   
-  if len(prompt) == 0:
+  if len(prompt) == 0 or len(prompt_en) == 0 :
     prompt = """Question: Can you help me please?
     Answer: Sure, I am here for you.
-    Question: """
+    Question: What do you do when you don't get what you want?"""
   
-  if language == 'en':
-    prompt = p_en + prompt + "\n" + "Answer: "
-  elif language == 'es':
+  #if language == 'en':
+  prompt = p_en + prompt_en + "\n" + "Answer: "
+  solution_en = query(prompt, 'en')
+  if language == 'es':
     prompt = p_es + prompt + "\n" + "Responder: "
+    solution = query(prompt, 'es')
   elif language == 'fr':
     prompt = p_fr + prompt + "\n" + "Réponse: "
+    solution = query(prompt, 'fr')
     
+  return solution, solution_en
+
+# Bloom API Request
+def query(prompt, language):
   json_ = {"inputs": prompt,
             "parameters":
             {
@@ -150,11 +157,12 @@ def lang_model_response(prompt, language):
   return solution
 
 # Coqui - Text-to-Speech
-def tts(text, language):
+def tts(text, text_en, language):
   print(f"Inside tts - language is : {language}")
   coqui_langs = ['en' ,'es' ,'fr' ,'de' ,'pl' ,'uk' ,'ro' ,'hu' ,'bg' ,'nl' ,'fi' ,'sl' ,'lv' ,'ga']
-  if language not in coqui_langs:
+  if language =='en' or language not in coqui_langs:
     language = 'en'
+    text = text_en
   with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
       coquiTTS.get_tts(text, fp, speaker = {"language" : language})
       return fp.name
@@ -168,18 +176,19 @@ with demo:
   with gr.Row():
     with gr.Column():
       in_audio = gr.Audio(source="microphone",  type="filepath", label='Record your voice here')  #type='filepath'
-      b1 = gr.Button("Whisper") #- Bloom - Coqui pipeline)")
+      b1 = gr.Button("Whisper") #- Bloom - Coqui pipeline
       out_transcript = gr.Textbox(label= 'As is Transcript using OpenAI Whisper')
       out_translation_en = gr.Textbox(label= 'English Translation of audio using OpenAI Whisper')
       out_lang = gr.Textbox(visible=False)
     with gr.Column():
-      out_audio = gr.Audio(label='AI response in Audio form in your preferred language')  
+      b2 = gr.Button("Bloom") #-- Coqui pipeline
       out_generated_text = gr.Textbox(label= 'AI response to your query in your preferred language using Bloom! ')
       out_generated_text_en = gr.Textbox(label= 'AI response to your query in English using Bloom! ')
-    
-      b1.click(whisper_stt, inputs=[in_audio], outputs=[out_transcript, out_translation_en, out_lang])
-      b2.click(
+      b3 = gr.Button("CoquiTTS") #-- pipeline complets
+      out_audio = gr.Audio(label='AI response in Audio form in your preferred language')  
       
-      b1.click(driver_fun,inputs=[in_audio], outputs=[out_transcript, out_translation_en, out_generated_text,out_generated_text_en, out_audio]) 
+      b1.click(whisper_stt, inputs=[in_audio], outputs=[out_transcript, out_translation_en, out_lang])
+      b2.click(lang_model_response, inputs=[out_transcript, out_translation_en, out_lang], outputs=[out_generated_text,out_generated_text_en])
+      b3.click(tts,inputs=[out_generated_text,out_generated_text_en,out_lang], outputs=[out_audio]) 
     
 demo.launch(enable_queue=True, debug=True)