Update app.py

app.py

@@ -1,20 +1,19 @@
-import os
 import base64
+import os
 from flask import Flask, request, jsonify
 from pydub import AudioSegment
 import whisper
 from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
 
-# Set environment variables for Hugging Face cache
-os.environ["TRANSFORMERS_CACHE"] = "/app/cache"
-os.environ["HF_HOME"] = "/app/cache"
+# Define cache directory
+os.environ['HF_HOME'] = '/app/cache'
 
-# Load the Whisper model
-whisper_model = whisper.load_model("base")
+# Load the Whisper model with a specified cache directory
+whisper_model = whisper.load_model("base", download_root="/app/cache")
 
 # Load the translation model and tokenizer
-tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M")
-translation_model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
+tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M", cache_dir="/app/cache")
+translation_model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M", cache_dir="/app/cache")
 
 def preprocess_audio(audio_path):
     """Convert audio to 16kHz mono WAV format."""
@@ -67,7 +66,7 @@ def translate():
     data = request.json
     if 'audio' not in data or 'source_lang' not in data or 'target_lang' not in data:
         return jsonify({"error": "Invalid request format"}), 400
-
+    
     audio_base64 = data['audio']
     source_lang = data['source_lang']
     target_lang = data['target_lang']