Update app.py

app.py

@@ -1,46 +1,116 @@
+# import gradio as gr
+# import whisper
+# import os
+# from groq import Groq
+
+# # 🔐 Get Groq API key securely from Hugging Face Secrets
+# GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+# groq_client = Groq(api_key=GROQ_API_KEY)
+# MODEL_NAME = "llama3-8b-8192"
+
+# # 🎙 Load Whisper
+# transcriber = whisper.load_model("base")
+
+# def transcribe_and_summarize(audio):
+#     # Step 1: Transcribe + Detect Language
+#     result = transcriber.transcribe(audio)
+#     transcript = result["text"]
+#     detected_lang = result["language"]
+
+#     # Step 2: Summarize in the same language
+#     if detected_lang == "en":
+#         system_prompt = "You are an expert English summarizer."
+#         user_prompt = f"Please summarize the following English text:\n\n{transcript}"
+#     elif detected_lang == "ur":
+#         system_prompt = "آپ ایک ماہر خلاصہ نگار ہیں جو اردو میں خلاصہ فراہم کرتے ہیں۔"
+#         user_prompt = f"براہ کرم مندرجہ ذیل اردو متن کا خلاصہ فراہم کریں:\n\n{transcript}"
+#     else:
+#         system_prompt = "You are a helpful summarizer."
+#         user_prompt = f"Summarize this text:\n\n{transcript}"
+
+#     response = groq_client.chat.completions.create(
+#         model=MODEL_NAME,
+#         messages=[
+#             {"role": "system", "content": system_prompt},
+#             {"role": "user", "content": user_prompt}
+#         ]
+#     )
+#     summary = response.choices[0].message.content.strip()
+
+#     lang_label = "English" if detected_lang == "en" else "Urdu" if detected_lang == "ur" else detected_lang.upper()
+
+#     return f"[{lang_label}] {transcript}", f"[{lang_label}] {summary}"
+
+# demo = gr.Interface(
+#     fn=transcribe_and_summarize,
+#     inputs=gr.Audio(type="filepath", label="🎧 Upload Audio (English or Urdu)"),
+#     outputs=[
+#         gr.Textbox(label="📝 Transcript"),
+#         gr.Textbox(label="🧠 Summary")
+#     ],
+#     title="🗣️ Multilingual Audio Summarizer",
+#     description="Upload English or Urdu audio. The app transcribes and summarizes in the same language using Whisper + Groq."
+# )
+
+# demo.launch()
+
+
 import gradio as gr
-import whisper
 import os
 from groq import Groq
+from transformers import pipeline
+import torchaudio
 
-# 🔐 Get Groq API key securely from Hugging Face Secrets
+# 🔐 Groq API key from environment
 GROQ_API_KEY = os.getenv("GROQ_API_KEY")
 groq_client = Groq(api_key=GROQ_API_KEY)
 MODEL_NAME = "llama3-8b-8192"
 
-# 🎙 Load Whisper
-transcriber = whisper.load_model("base")
-
-def transcribe_and_summarize(audio):
-    # Step 1: Transcribe + Detect Language
-    result = transcriber.transcribe(audio)
-    transcript = result["text"]
-    detected_lang = result["language"]
-
-    # Step 2: Summarize in the same language
-    if detected_lang == "en":
-        system_prompt = "You are an expert English summarizer."
-        user_prompt = f"Please summarize the following English text:\n\n{transcript}"
-    elif detected_lang == "ur":
-        system_prompt = "آپ ایک ماہر خلاصہ نگار ہیں جو اردو میں خلاصہ فراہم کرتے ہیں۔"
-        user_prompt = f"براہ کرم مندرجہ ذیل اردو متن کا خلاصہ فراہم کریں:\n\n{transcript}"
-    else:
-        system_prompt = "You are a helpful summarizer."
-        user_prompt = f"Summarize this text:\n\n{transcript}"
-
-    response = groq_client.chat.completions.create(
-        model=MODEL_NAME,
-        messages=[
-            {"role": "system", "content": system_prompt},
-            {"role": "user", "content": user_prompt}
-        ]
-    )
-    summary = response.choices[0].message.content.strip()
-
-    lang_label = "English" if detected_lang == "en" else "Urdu" if detected_lang == "ur" else detected_lang.upper()
-
-    return f"[{lang_label}] {transcript}", f"[{lang_label}] {summary}"
+# ⚡ Use fast Whisper Tiny ASR model from Hugging Face
+asr = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
+
+# Limit audio duration to 5 mins (300 seconds)
+MAX_DURATION = 300
+
+def transcribe_and_summarize(audio_path):
+    try:
+        # Check audio duration
+        info = torchaudio.info(audio_path)
+        duration = info.num_frames / info.sample_rate
+        if duration > MAX_DURATION:
+            return "❌ Audio too long. Please upload audio under 5 minutes.", ""
+
+        # Transcribe using Whisper-tiny
+        result = asr(audio_path)
+        transcript = result["text"]
+        detected_lang = result.get("language", "en")  # fallback to 'en' if missing
+
+        # Summarize with Groq
+        if detected_lang == "en":
+            system_prompt = "You are an expert English summarizer."
+            user_prompt = f"Please summarize the following English text:\n\n{transcript}"
+        elif detected_lang == "ur":
+            system_prompt = "آپ ایک ماہر خلاصہ نگار ہیں جو اردو میں خلاصہ فراہم کرتے ہیں۔"
+            user_prompt = f"براہ کرم مندرجہ ذیل اردو متن کا خلاصہ فراہم کریں:\n\n{transcript}"
+        else:
+            system_prompt = "You are a helpful summarizer."
+            user_prompt = f"Summarize this text:\n\n{transcript}"
+
+        response = groq_client.chat.completions.create(
+            model=MODEL_NAME,
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": user_prompt}
+            ]
+        )
+        summary = response.choices[0].message.content.strip()
+        lang_label = "English" if detected_lang == "en" else "Urdu" if detected_lang == "ur" else detected_lang.upper()
+        return f"[{lang_label}] {transcript}", f"[{lang_label}] {summary}"
+
+    except Exception as e:
+        return "❌ Error processing audio. Please try again.", str(e)
 
+# Gradio UI
 demo = gr.Interface(
     fn=transcribe_and_summarize,
     inputs=gr.Audio(type="filepath", label="🎧 Upload Audio (English or Urdu)"),
@@ -49,7 +119,7 @@ demo = gr.Interface(
         gr.Textbox(label="🧠 Summary")
     ],
     title="🗣️ Multilingual Audio Summarizer",
-    description="Upload English or Urdu audio. The app transcribes and summarizes in the same language using Whisper + Groq."
+    description="Upload English or Urdu audio (max 5 minutes). The app transcribes and summarizes using Whisper-Tiny + Groq."
 )
 
 demo.launch()