Update app.py

app.py

@@ -1,7 +1,7 @@
 import gradio as gr
 import random
 import torch
-from transformers import AutoConfig
+from transformers import AutoConfig, AutoTokenizer, AutoModelWithLMHead
 from transformers import GPT2Tokenizer, GPT2LMHeadModel
 from itertools import chain
 
@@ -14,7 +14,17 @@ import numpy as np
 from TTS.utils.manage import ModelManager
 from TTS.utils.synthesizer import Synthesizer
 
+emotion_tokenizer = AutoTokenizer.from_pretrained("mrm8488/t5-base-finetuned-emotion")
+emotion_model = AutoModelWithLMHead.from_pretrained("mrm8488/t5-base-finetuned-emotion")
 
+def get_emotion(text):
+  input_ids = tokenizer.encode(text + '</s>', return_tensors='pt')
+  output = model.generate(input_ids=input_ids,max_length=2)
+  dec = [tokenizer.decode(ids) for ids in output]
+  label = dec[0]
+  return label.split()[1]
+
+    
 config = AutoConfig.from_pretrained('gorkemgoknar/gpt2chatbotenglish')
 model = GPT2LMHeadModel.from_pretrained('gorkemgoknar/gpt2chatbotenglish', config=config)
 
@@ -22,6 +32,7 @@ tokenizer = GPT2Tokenizer.from_pretrained('gorkemgoknar/gpt2chatbotenglish')
 tokenizer.model_max_length = 1024
 
 
+
 #Dynamic Temperature 
 #See experiment https://www.linkedin.com/pulse/ai-goes-job-interview-g%25C3%25B6rkem-g%25C3%25B6knar
 
@@ -119,7 +130,8 @@ def greet(character,your_voice,message,history):
   os.system('tts --text "'+response+'" --model_name tts_models/multilingual/multi-dataset/your_tts --speaker_wav '+speaker_wav+' --language_idx "en"') 
 
   history["message_history"].append((message, response))
-  
+
+  emotion = get_emotion(response)
   
   html = "<div class='chatbot'>"
   for user_msg, resp_msg in history["message_history"]:
@@ -127,7 +139,7 @@ def greet(character,your_voice,message,history):
       html += f"<div class='resp_msg'>{character}: {resp_msg}</div>"
   html += "</div>"
 
-  return "tts_output.wav",html,history
+  return "tts_output.wav",emotion, html,history
   
   
 
@@ -157,7 +169,7 @@ examples=[['Gandalf','Hello','dragon.wav']]
 history =   {"character": "None", "message_history" : [] }
 interface= gr.Interface(fn=greet, 
                         inputs=[gr.inputs.Dropdown(personality_choices),gr.inputs.Audio(source="microphone", type="filepath") ,"text", "state"], 
-                        outputs=[gr.outputs.Audio(type="file"),"html","state"],      
+                        outputs=[gr.outputs.Audio(type="file"),"text","html","state"],      
                         css=css, title=title, description=description,article=article )