Update app.py

app.py

@@ -303,6 +303,111 @@ class GradioInference:
  wordcloud_image,
  )
 
+ 
+ def from_article(self, article, progress=gr.Progress()):
+ """
+ Call the Gradio Inference python class.
+ Uses it directly the Whisper model to perform Automatic Speech Recognition (i.e Speech-to-Text).
+ Once the function has the transcription of the video it proccess it to obtain:
+ - Summary: using Facebook's BART transformer.
+ - KeyWords: using VoiceLabT5 keyword extractor.
+ - Sentiment Analysis: using Hugging Face's default sentiment classifier
+ - WordCloud: using the wordcloud python library.
+ """
+ progress(0, desc="Starting analysis")
+
+ progress(0.30, desc="Summarizing")
+ 
+ # Perform summarization on the transcription
+ transcription_summary = self.bart_summarizer(
+ results["text"], max_length=150, min_length=30, do_sample=False, truncation=True
+ )
+ 
+ #### Resumen multilingue
+ WHITESPACE_HANDLER = lambda k: re.sub('\s+', ' ', re.sub('\n+', ' ', k.strip()))
+ 
+ input_ids_sum = self.mt5_tokenizer(
+ [WHITESPACE_HANDLER(results["text"])],
+ return_tensors="pt",
+ padding="max_length",
+ truncation=True,
+ max_length=512
+ )["input_ids"]
+ 
+ output_ids_sum = self.mt5_model.generate(
+ input_ids=input_ids_sum,
+ max_length=130,
+ no_repeat_ngram_size=2,
+ num_beams=4
+ )[0]
+ 
+ summary = self.mt5_tokenizer.decode(
+ output_ids_sum,
+ skip_special_tokens=True,
+ clean_up_tokenization_spaces=False
+ )
+ #### Fin resumen multilingue
+
+ progress(0.60, desc="Extracting Keywords")
+ 
+ # Extract keywords using VoiceLabT5
+ task_prefix = "Keywords: "
+ input_sequence = task_prefix + results["text"]
+ 
+ input_ids = self.keyword_tokenizer(
+ input_sequence, 
+ return_tensors="pt", 
+ truncation=False
+ ).input_ids
+ 
+ output = self.keyword_model.generate(
+ input_ids, 
+ no_repeat_ngram_size=3, 
+ num_beams=4
+ )
+ predicted = self.keyword_tokenizer.decode(output[0], skip_special_tokens=True)
+ keywords = [x.strip() for x in predicted.split(",") if x.strip()]
+ formatted_keywords = "\n".join([f"• {keyword}" for keyword in keywords])
+
+ progress(0.80, desc="Extracting Sentiment")
+
+ # Define a dictionary to map labels to emojis
+ sentiment_emojis = {
+ "positive": "Positive 👍🏼",
+ "negative": "Negative 👎🏼",
+ "neutral": "Neutral 😶",
+ }
+ 
+ # Sentiment label 
+ label = self.classifier(summary)[0]["label"]
+
+ # Format the label with emojis
+ formatted_sentiment = sentiment_emojis.get(label, label)
+ 
+ progress(0.90, desc="Generating Wordcloud")
+ # WordCloud object
+ wordcloud = WordCloud(colormap = "Oranges").generate(
+ results["text"]
+ )
+ wordcloud_image = wordcloud.to_image()
+
+ if lang == "english" or lang == "none":
+ return (
+ results["text"],
+ transcription_summary[0]["summary_text"],
+ formatted_keywords,
+ formatted_sentiment,
+ wordcloud_image,
+ )
+ else:
+ return (
+ results["text"],
+ summary,
+ formatted_keywords,
+ formatted_sentiment,
+ wordcloud_image,
+ )
+
 
 gio = GradioInference()
 title = "YouTube Insights"
@@ -409,6 +514,44 @@ with block as demo:
  outputs=[text, summary, keywords, label, wordcloud_image],
  )
 
+ with gr.Tab("From Article 📋"):
+ with gr.Box():
+ 
+ with gr.Row().style(equal_height=True):
+ size = gr.Dropdown(
+ label="Model Size", choices=gio.sizes, value="base"
+ )
+ lang = gr.Dropdown(
+ label="Language (Optional)", choices=gio.langs, value="none"
+ )
+ 
+ with gr.Row().style(equal_height=True):
+ article = gr.Textbox(
+ label="Transcription",
+ placeholder="Paste your text...",
+ lines=10,
+ ).style(show_copy_button=True, container=False)
+ 
+ with gr.Row().style(equal_height=True):
+ summary = gr.Textbox(
+ label="Summary", placeholder="Summary Output", lines=5
+ )
+ keywords = gr.Textbox(
+ label="Keywords", placeholder="Keywords Output", lines=5
+ )
+ label = gr.Label(label="Sentiment Analysis")
+ wordcloud_image = gr.Image(label="WordCloud")
+ 
+ with gr.Row().style(equal_height=True):
+ clear = gr.ClearButton([audio_file,text, summary, keywords, label, wordcloud_image], scale=1, value="Clear 🗑️")
+ btn = gr.Button(
+ "Get audio insights 🔎", variant="primary", scale=1
+ )
+ btn.click(
+ gio.from_audio_input,
+ inputs=[lang, size, article],
+ outputs=[summary, keywords, label, wordcloud_image],
+ )
 
 with block:
  gr.Markdown("### Video Examples")