lelo/bethaus-app
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

cuda with cpu fallback

Browse Source
This commit is contained in:
lelo 2025-06-01 09:35:42 +00:00
parent b2e61129d8
commit 05cbd5efa1
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
transcribe_all.py
View File

@ -1,6 +1,7 @@
import os import os
import sys import sys
import time import time
import torch
import whisper import whisper
import concurrent.futures import concurrent.futures
import json import json
@ -18,7 +19,6 @@ with open("transcription_config.yml", "r", encoding="utf-8") as file:
settings = yaml.safe_load(file) settings = yaml.safe_load(file)
folder_list = settings.get("folder_list") folder_list = settings.get("folder_list")
model_name = settings.get("model_name") model_name = settings.get("model_name")
device = settings.get("device")
def load_audio_librosa(path: str, sr: int = 16_000) -> np.ndarray: def load_audio_librosa(path: str, sr: int = 16_000) -> np.ndarray:
audio, orig_sr = librosa.load(path, sr=sr) # load + resample to 16 kHz audio, orig_sr = librosa.load(path, sr=sr) # load + resample to 16 kHz
@ -222,7 +222,10 @@ def process_folder(root_folder):
else: else:
print(f"Checked {checked_files} files. Start to transcribe {len(valid_files)} files.") print(f"Checked {checked_files} files. Start to transcribe {len(valid_files)} files.")
print("Loading Whisper model...") # Choose “cuda” if available, otherwise “cpu”
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Loading Whisper model on {device}…")
model = whisper.load_model(model_name, device=device) model = whisper.load_model(model_name, device=device)
# Use a thread pool to pre-load files concurrently. # Use a thread pool to pre-load files concurrently.