Update denoiser.py

denoiser.py

@@ -1,18 +1,19 @@
 """
 Department 1 — Cloud Denoiser (Cleanvoice API)
-The "Vacuum-Sealed" Podcast Sound
+The "Vacuum-Sealed" Podcast Sound (MP3 Output)
 """
 import os
 import requests
 import tempfile
 import logging
 import re
+import subprocess
 
 logger = logging.getLogger(__name__)
 
 class Denoiser:
     def __init__(self):
-        # ⚠️ PASTE YOUR API KEY RIGHT HERE! ⚠️
+        # ⚠️ API KEY LOADED
         self.api_key = "cvk_Xnmz7ZPXMRg.CcE3YGry3IzkCBWQKGvomBfd9hmC9L-aAvSj-2iNfyw"
         print("[Denoiser] ☁️ Cleanvoice Cloud API loaded")
 
@@ -27,9 +28,6 @@ class Denoiser:
             from cleanvoice import Cleanvoice
         except ImportError:
             raise RuntimeError("Please add cleanvoice-sdk to your requirements.txt")
-            
-        if self.api_key == "YOUR_CLEANVOICE_API_KEY_HERE":
-            raise ValueError("You need to paste your Cleanvoice API key into denoiser.py!")
 
         cv = Cleanvoice({'api_key': self.api_key})
         
@@ -46,28 +44,40 @@ class Denoiser:
             breath=remove_breaths
         )
         
-        logger.info(f"[Denoiser] Cloud processing complete. Downloading result...")
-        
-        fd, final_audio_path = tempfile.mkstemp(suffix="_cv_cleaned.wav", dir=out_dir)
-        os.close(fd)
+        logger.info(f"[Denoiser] Cloud processing complete. Downloading and converting to MP3...")
         
-        # Download the processed studio-quality audio
+        # 1. Download the raw audio from Cleanvoice
         audio_data = requests.get(result.audio.url).content
-        with open(final_audio_path, "wb") as f:
+        
+        fd, temp_dl_path = tempfile.mkstemp(suffix=".tmp", dir=out_dir)
+        os.close(fd)
+        with open(temp_dl_path, "wb") as f:
             f.write(audio_data)
             
+        # 2. Convert directly to high-quality MP3 using ffmpeg
+        final_mp3_path = os.path.join(out_dir, "cleanvoice_studio.mp3")
+        subprocess.run([
+            "ffmpeg", "-y", "-i", temp_dl_path, 
+            "-codec:a", "libmp3lame", "-qscale:a", "2", 
+            final_mp3_path
+        ], capture_output=True)
+        
+        # 3. Clean up the temporary download
+        try:
+            os.remove(temp_dl_path)
+        except OSError:
+            pass
+            
         return {
-            'audio_path': final_audio_path,
+            'audio_path': final_mp3_path,
             'stats': {
-                'noise_method': "Cleanvoice API (Studio Sound)",
+                'noise_method': "Cleanvoice API (Studio MP3)",
                 'processing_sec': 0,
                 'fillers_removed': "Cleanvoice",
                 'stutters_removed': "Cleanvoice"
             }
         }
         
-    # We keep this dummy method so your UI doesn't crash!
-    # Cleanvoice cuts the audio, but we still need to clean the text transcript.
     def clean_transcript_fillers(self, transcript: str) -> str:
         FILLER_WORDS = {
             "um", "umm", "uh", "hmm", "like", "basically", "literally", 
@@ -80,12 +90,12 @@ class Denoiser:
             if clean_word not in FILLER_WORDS:
                 result.append(w)
         return " ".join(result)
-# ══════════════════════════════════════════════════════════════════
+
+    # ══════════════════════════════════════════════════════════════════
     # DUMMY METHODS (To prevent app.py runtime crashes)
-    # Cleanvoice already cut these in the cloud, so we just return the audio as-is.
     # ══════════════════════════════════════════════════════════════════
     def _remove_fillers(self, audio, sr, segments):
         return audio, 0
         
     def _remove_stutters(self, audio, sr, segments):
-        return audio, 0        
+        return audio, 0