reduced spaces duration

app.py

@@ -10,7 +10,6 @@ if torch.cuda.is_available():
 else:
     print("No CUDA GPUs available. Running on CPU.")
 
-# @spaces.GPU(duration=180)
 def transcribe_audio(audio_file, translate, model_size):
     language_segments, final_segments = process_audio(audio_file, translate=translate, model_size=model_size)
     

audio_processing.py

@@ -16,11 +16,11 @@ OVERLAP=0
 import whisperx
 import torch
 import numpy as np
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-import spaces
 
 
+logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+logger = logging.getLogger(__name__)
+import spaces
 
 
 def preprocess_audio(audio, chunk_size=CHUNK_LENGTH*16000, overlap=OVERLAP*16000):  # 2 seconds overlap
@@ -32,7 +32,7 @@ def preprocess_audio(audio, chunk_size=CHUNK_LENGTH*16000, overlap=OVERLAP*16000
         chunks.append(chunk)
     return chunks
 
-@spaces.GPU(duration=10)
+@spaces.GPU()
 def process_audio(audio_file, translate=False, model_size="small"):
     start_time = time.time()
     
@@ -113,7 +113,7 @@ def process_audio(audio_file, translate=False, model_size="small"):
         logger.error(f"An error occurred during audio processing: {str(e)}")
         raise
 
-def merge_nearby_segments(segments, time_threshold=0.5, similarity_threshold=0.7):
+def merge_nearby_segments(segments, time_threshold=0.5, similarity_threshold=0.9):
     merged = []
     for segment in segments:
         if not merged or segment['start'] - merged[-1]['end'] > time_threshold:

requirements.txt

@@ -19,5 +19,3 @@ nvidia-cublas-cu11
 torchvision
 cdifflib
 pydub
-
-