update

app.py

@@ -10,6 +10,10 @@ from datasets import load_dataset
 import random
 
 dataset = load_dataset("ysharma/short_jokes", split="train")
+filtered_dataset = dataset.filter(
+    lambda x: (True not in [nsfw in x["Joke"].lower() for nsfw in ["warning", "fuck", "dead", "nsfw","69", "sex"]]) 
+    )
+
 
 # Model 2: Sentence Transformer
 API_URL = "https://api-inference.huggingface.co/models/sentence-transformers/msmarco-distilbert-base-tas-b"
@@ -47,7 +51,7 @@ def driver_fun(audio) :
     lower_limit = random_val - 5000
     upper_limit = random_val 
   print(f"lower_limit : upper_limit = {lower_limit} : {upper_limit}")  
-  dataset_subset = dataset['Joke'][lower_limit : upper_limit]
+  dataset_subset = filtered_dataset['Joke'][lower_limit : upper_limit]
   data = query({"inputs": {"source_sentence": "That is a happy person","sentences": dataset_subset} } )
   if 'error' in data:
     print(f"Error is : {data}")
@@ -106,7 +110,7 @@ demo = gr.Blocks()
 with demo:
   gr.Markdown("<h1><center>AI Assistant - Voice to Joke</center></h1>")
   gr.Markdown(
-        """<center>This is still a work in porgress</center><br><br><br>Model pipeline consisting of - <br>- [**Whisper**](https://github.com/openai/whisper) for Speech-to-text, <br>- [**CoquiTTS**](https://huggingface.co/coqui)  for Text-To-Speech. <br>- Front end is built using [**Gradio Block API**](https://gradio.app/docs/#blocks).<br><be><u>Ask Whisper for a joke about anything you would wish</u>.<br>If you want to reuse the App, simply click on the small cross button in the top right corner of your voice record panel, and then press record again!
+        """This is still a work in porgress<br><br>Ask Whisper for a joke about anything you would wish.
         """)
   with gr.Row():
     with gr.Column(): 
@@ -120,5 +124,9 @@ with demo:
       #out_generated_text_en = gr.Textbox(label= 'AI response to your query in English using Bloom! ')
     
       b1.click(driver_fun,inputs=[in_audio], outputs=[out_transcript, out_generated_joke, out_audio]) #out_translation_en, out_generated_text,out_generated_text_en, 
-    
+  with gr.Row():
+    gr.Markdown(
+        """Model pipeline consisting of - <br>- [**Whisper**](https://github.com/openai/whisper) for Speech-to-text, <br>- [**CoquiTTS**](https://huggingface.co/coqui)  for Text-To-Speech. <br>- Front end is built using [**Gradio Block API**](https://gradio.app/docs/#blocks).<br><be>If you want to reuse the App, simply click on the small cross button in the top right corner of your voice record panel, and then press record again!
+        """)
+  
 demo.launch(enable_queue=True, debug=True)