Update app.py

app.py

@@ -1,44 +1,76 @@
-def run_pipeline(audio_input, remove_fillers, remove_stutters, remove_silences, remove_breaths, remove_mouth, target_lang):
-    if not audio_input:
-        return None, "No audio detected. Please upload a file.", ""
+import gradio as gr
+import os
+import subprocess
+import logging
+from datetime import datetime
+
+from denoiser import Denoiser
+from transcriber import Transcriber
+from translator import Translator
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
 
-    # Create job directory
-    job_id = datetime.now().strftime("%Y%m%d_%H%M%S")
-    output_dir = f"outputs/{job_id}"
-    os.makedirs(output_dir, exist_ok=True)
+denoiser = Denoiser()
+transcriber = Transcriber()
+translator = Translator()
+
+def run_pipeline(audio_input, remove_fillers, remove_stutters, remove_silences, remove_breaths, remove_mouth, target_lang):
+    if not audio_input: return None, "Please upload a file.", ""
+    
+    out_dir = f"outputs/{datetime.now().strftime('%H%M%S')}"
+    os.makedirs(out_dir, exist_ok=True)
     
     try:
-        # 1. DENOISE & CLEAN
-        # The new 'process' method handles everything via the Cleanvoice API
-        logger.info("[Pipeline] Step 1: Denoising & Studio Processing...")
+        # 1. Clean Audio (API + WhatsApp pre-conversion)
         res = denoiser.process(
-            audio_input, output_dir,
+            audio_input, out_dir, 
             fillers=remove_fillers, 
             stutters=remove_stutters,
-            long_silences=remove_silences, 
+            long_silences=remove_silences,
             breath=remove_breaths,
             mouth_sounds=remove_mouth
         )
         
-        # 2. TRANSCRIBE
-        logger.info("[Pipeline] Step 2: Transcribing...")
-        # Use the processed audio path returned by the denoiser
-        transcript_data = transcriber.transcribe(res['audio_path'])
-        text = transcript_data['text']
+        # 2. Transcribe
+        transcript = transcriber.transcribe(res['audio_path'])['text']
         
-        # 3. TRANSLATE
-        logger.info(f"[Pipeline] Step 3: Translating to {target_lang}...")
-        translated = translator.translate(text, target_lang)
+        # 3. Translate
+        translated = translator.translate(transcript, target_lang)
 
-        # 4. FINAL EXPORT: Conversion to MP3
-        final_mp3 = os.path.join(output_dir, "ClearWave_Studio_Final.mp3")
+        # 4. Final MP3 Export
+        final_mp3 = os.path.join(out_dir, "ClearWave_Final.mp3")
         subprocess.run([
             "ffmpeg", "-y", "-i", res['audio_path'],
             "-codec:a", "libmp3lame", "-qscale:a", "2", final_mp3
         ], check=True)
 
-        return final_mp3, text, translated
-
+        return final_mp3, transcript, translated
     except Exception as e:
-        logger.error(f"Pipeline Error: {e}")
-        return None, f"Error: {str(e)}", ""
+        logger.error(f"Pipeline crashed: {e}")
+        return None, f"Error: {e}", ""
+
+with gr.Blocks(title="ClearWave AI Studio") as demo:
+    gr.Markdown("# 🎙️ ClearWave AI Studio")
+    gr.Markdown("Transform WhatsApp audio into studio-quality MP3s.")
+    
+    with gr.Row():
+        with gr.Column():
+            # Gradio 6.0: type="filepath" is essential for FFmpeg to find the file
+            audio_in = gr.Audio(label="Upload Audio", type="filepath")
+            with gr.Row():
+                f = gr.Checkbox(label="Fillers", value=True)
+                s = gr.Checkbox(label="Stutters", value=True)
+                z = gr.Checkbox(label="Silences", value=True)
+            lang = gr.Dropdown(choices=["Hindi", "Telugu", "Spanish"], value="Hindi", label="Translation")
+            btn = gr.Button("✨ Process to Studio MP3", variant="primary")
+        
+        with gr.Column():
+            audio_out = gr.Audio(label="Cleaned MP3 Result")
+            # Removed show_copy_button to fix Gradio 6.0 error
+            t_out = gr.Textbox(label="Transcript", lines=5)
+            tr_out = gr.Textbox(label="Translation", lines=5)
+
+    btn.click(run_pipeline, [audio_in, f, s, z, f, f, lang], [audio_out, t_out, tr_out])
+
+demo.launch()