Update app.py

app.py

@@ -1,161 +1,137 @@
-# =========================
-# SmartTranscribe - Updated Version (For Hugging Face Spaces)
-# =========================
-
 import os
 import gradio as gr
 import requests
-from groq import Groq
-from datetime import datetime
-from pathlib import Path
 import tempfile
+import yt_dlp
+from groq import Groq
 from huggingface_hub import InferenceClient
 
-# -------------------------
-# Environment Variables
-# -------------------------
-GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
-HUGGINGFACE_API_TOKEN = os.environ.get("HUGGINGFACE_API_TOKEN")
+# ----------------------------
+# ✅ Environment Variables
+# ----------------------------
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+HUGGINGFACE_API_TOKEN = os.getenv("HUGGINGFACE_API_TOKEN")
 
 if not GROQ_API_KEY or not HUGGINGFACE_API_TOKEN:
-    raise EnvironmentError(
-        "Environment variables GROQ_API_KEY and HUGGINGFACE_API_TOKEN must be set."
-    )
+    raise EnvironmentError("Please set GROQ_API_KEY and HUGGINGFACE_API_TOKEN in Hugging Face settings.")
 
-# Initialize Groq Client
+# ----------------------------
+# ✅ Initialize Clients
+# ----------------------------
 groq_client = Groq(api_key=GROQ_API_KEY)
-
-# Initialize Hugging Face Inference Client
 hf_client = InferenceClient(token=HUGGINGFACE_API_TOKEN)
 
-# -------------------------
-# Utility Functions
-# -------------------------
+# ----------------------------
+# ✅ Download YouTube Audio
+# ----------------------------
+def download_youtube_audio(youtube_url):
+    try:
+        with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as tmp_file:
+            ydl_opts = {
+                'format': 'bestaudio/best',
+                'outtmpl': tmp_file.name,
+                'quiet': True,
+                'postprocessors': [{
+                    'key': 'FFmpegExtractAudio',
+                    'preferredcodec': 'mp3',
+                    'preferredquality': '192',
+                }],
+            }
+            with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+                ydl.download([youtube_url])
+            return tmp_file.name
+    except Exception as e:
+        return f"❌ Error downloading audio: {e}"
 
-def transcribe_audio(audio_path, language=None):
-    """
-    Transcribe Urdu or English audio using Whisper Large-v3 Turbo on Groq.
-    Auto-detects language and returns cleaned text.
-    """
+# ----------------------------
+# ✅ Transcribe with Groq Whisper
+# ----------------------------
+def transcribe_audio(audio_path):
     try:
-        with open(audio_path, "rb") as audio_file:
-            response = groq_client.audio.transcriptions.create(
-                model="whisper-large-v3-turbo",
-                file=audio_file,
-                response_format="text"
+        with open(audio_path, "rb") as f:
+            transcription = groq_client.audio.transcriptions.create(
+                model="whisper-large-v3",
+                file=f
             )
-
-        transcript = response.strip()
-
-        # Urdu-English post-processing (convert English words in Urdu audio into Urdu script)
-        transcript = normalize_transcription(transcript)
-
-        return transcript
-
+        return transcription.text
     except Exception as e:
-        return f"❌ Error during transcription: {str(e)}"
-
-
-def normalize_transcription(text):
-    """
-    Basic normalization for Urdu-English blend.
-    You can enhance this later with a proper transliteration module.
-    """
-    replacements = {
-        "school": "اسکول",
-        "teacher": "ٹیچر",
-        "student": "سٹوڈنٹ",
-        "education": "ایجوکیشن",
-        "university": "یونیورسٹی",
-        "computer": "کمپیوٹر",
-        "mobile": "موبائل",
-        "class": "کلاس",
-    }
-    for eng, urdu in replacements.items():
-        text = text.replace(eng, urdu)
-    return text
-
-
-def summarize_text(text):
-    """
-    Summarize text using openai/gpt-oss-120b model from Hugging Face.
-    """
+        return f"❌ Error during transcription: {e}"
+
+# ----------------------------
+# ✅ Summarize in English or Urdu
+# ----------------------------
+def summarize_text(text, lang):
     try:
-        summary_prompt = f"Summarize the following text in the same language (Urdu or English):\n\n{text}\n\nSummary:"
-        response = hf_client.text_generation(
-            model="openai/gpt-oss-120b",
-            inputs=summary_prompt,
+        if lang == "English":
+            model = "facebook/bart-large-cnn"
+            prompt = f"Summarize the following text in English:\n\n{text}"
+        else:
+            model = "facebook/mbart-large-50-many-to-many-mmt"
+            prompt = f"مندرجہ ذیل انگریزی متن کا جامع اردو خلاصہ لکھیں:\n\n{text}"
+
+        output = hf_client.text_generation(
+            model=model,
+            prompt=prompt,
             max_new_tokens=250,
-            temperature=0.5,
+            temperature=0.7,
+            do_sample=False,
         )
-        return response.generated_text.strip()
+        return output
     except Exception as e:
-        return f"❌ Error during summarization: {str(e)}"
-
-
-def process_audio(audio_path):
-    """Main pipeline for transcription + summarization"""
-    if not audio_path:
-        return "⚠️ Please upload or record an audio/video file.", ""
-
+        return f"❌ Error during summarization: {e}"
+
+# ----------------------------
+# ✅ Main Function: YouTube or File
+# ----------------------------
+def process_input(youtube_url, audio_file, lang):
+    audio_path = None
+
+    # Step 1: Determine source
+    if youtube_url:
+        audio_path = download_youtube_audio(youtube_url)
+        if "❌" in audio_path:
+            return audio_path, "", ""
+    elif audio_file:
+        audio_path = audio_file
+    else:
+        return "❌ Please upload audio or paste a YouTube link.", "", ""
+
+    # Step 2: Transcription
     transcript = transcribe_audio(audio_path)
     if transcript.startswith("❌"):
-        return transcript, ""
-
-    summary = summarize_text(transcript)
-    return transcript, summary
+        return transcript, "", ""
 
+    # Step 3: Summarization
+    summary = summarize_text(transcript, lang)
 
-# -------------------------
-# Gradio Interface
-# -------------------------
+    return "✅ Transcription Completed!", transcript, summary
 
-with gr.Blocks(theme=gr.themes.Soft(), title="SmartTranscribe – Urdu & English AI Transcription") as app:
-    gr.Markdown(
-        """
-        # 🎙️ **SmartTranscribe**
-        **AI-Powered Urdu & English Transcription & Summarization App**
+# ----------------------------
+# ✅ Gradio UI
+# ----------------------------
+with gr.Blocks(title="🎧 Audio & YouTube Transcriber + Summarizer") as app:
+    gr.Markdown("## 🎧 English/Urdu Audio Summarizer\nUpload an audio file **or** paste a YouTube link below:")
 
-        Upload, record, or link your audio/video — the app will:
-        1. 🎧 Transcribe in the same language (Urdu/English)
-        2. 📝 Convert English words in Urdu speech into Urdu script
-        3. ✨ Generate a concise summary using `openai/gpt-oss-120b`
-        """
-    )
-
-    with gr.Tab("🎤 Upload or Record"):
-        audio_input = gr.Audio(
-            sources=["microphone", "upload"],  # ✅ Updated syntax
-            type="filepath",
-            label="Upload or Record audio/video"
+    with gr.Row():
+        youtube_link = gr.Textbox(label="📺 YouTube Link (optional)")
+        language_choice = gr.Dropdown(
+            ["English", "Urdu"],
+            label="🌐 Choose Summary Language",
+            value="English"
         )
 
-        transcribe_btn = gr.Button("🚀 Start Transcription")
-        transcript_output = gr.Textbox(label="📝 Transcribed Text", lines=10)
-        summary_output = gr.Textbox(label="📄 Summary", lines=8)
+    audio_input = gr.Audio(type="filepath", label="🎙️ Upload Audio File (optional)")
+    btn = gr.Button("🚀 Transcribe & Summarize")
 
-        transcribe_btn.click(
-            fn=process_audio,
-            inputs=audio_input,
-            outputs=[transcript_output, summary_output]
-        )
+    with gr.Row():
+        status = gr.Textbox(label="Status")
+    transcript_box = gr.Textbox(label="📝 Transcription", lines=8)
+    summary_box = gr.Textbox(label="🧩 Summary", lines=8)
 
-    with gr.Tab("ℹ️ About"):
-        gr.Markdown(
-            """
-            ### 💡 How it Works
-            - Uses **Groq + Whisper Large-v3 Turbo** for lightning-fast transcription  
-            - Post-processes mixed Urdu-English text into clean, grammatically correct Urdu  
-            - Generates summaries with **openai/gpt-oss-120b**  
-            
-            ### 🔒 Privacy
-            Your files and text are **not stored** after processing.  
-            All processing happens temporarily in memory.
-            """
-        )
+    btn.click(
+        fn=process_input,
+        inputs=[youtube_link, audio_input, language_choice],
+        outputs=[status, transcript_box, summary_box]
+    )
 
-# -------------------------
-# Launch App
-# -------------------------
-if __name__ == "__main__":
-    app.launch(server_name="0.0.0.0", server_port=7860)
+app.launch()