thi*_*way 4 python openai-whisper
tldr：我的假设是否正确torch.cuda.init()，device = "cuda"并且result = model.transcribe(etc)应该足以强制使用 GPU？
我检查了几个论坛帖子，但找不到解决方案。抱歉，如果这很愚蠢。我也在耳语 git 上发帖，但也许它不是耳语特定的。
简而言之，这是我的 python 脚本：
import whisper
import soundfile as sf
import torch
# specify the path to the input audio file
input_file = "H:\\path\\3minfile.WAV"
# specify the path to the output transcript file
output_file = "H:\\path\\transcript.txt"
# Cuda allows for the GPU to be used which is more optimized than the cpu
torch.cuda.init()
device = "cuda" # if torch.cuda.is_available() else "cpu"
# Load audio file
audio_data, sample_rate = sf.read(input_file, always_2d=True)
#load whisper model
model_size = "tiny"
print("loading model :", model_size)
model = whisper.load_model(model_size).to(device)
print(model_size, "model loaded")
# Initialize variables
results = []
language = "fr"
# Transcribe audio
with torch.cuda.device(device):
result = model.transcribe(audio_data, language=language, fp16=False, word_timestamps=True)
但是，它在最后一行返回以下错误，暗示它正在尝试在 cpu 上运行它：
RuntimeError: [在 C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\c10\core\impl\alloc_cpu.cpp:72 处执行失败] 数据。DefaultCPUAllocator：内存不足：您尝试分配 30623038517864 字节。
我正在使用 Jupyter，我检查了它使用的 pytorch 版本是 cuda/gpu 版本，而不是 cpu 锁定版本：
print(torch.__version__)
2.0.0+cu117
所以我真的不明白。pythorch库会不会有冲突？难道我做错了什么 ？transcribe() 函数确实使用 cpu 而不是 gpu 吗？
我正在使用 anaconda3，以下是conda list返回的内容，以防有帮助：
# Name Version Build Channel
_ipyw_jlab_nb_ext_conf 0.1.0 py39haa95532_0
alabaster 0.7.12 pyhd3eb1b0_0
anaconda 2022.10 py39_0
anaconda-client 1.11.0 py39haa95532_0
anaconda-navigator 2.3.1 py39haa95532_0
anaconda-project 0.11.1 py39haa95532_0
anyio 3.5.0 py39haa95532_0
appdirs 1.4.4 pyhd3eb1b0_0
argon2-cffi 21.3.0 pyhd3eb1b0_0
argon2-cffi-bindings 21.2.0 py39h2bbff1b_0
arrow 1.2.2 pyhd3eb1b0_0
astroid 2.11.7 py39haa95532_0
astropy 5.1 py39h080aedc_0
atomicwrites 1.4.0 py_0
attrs 21.4.0 pyhd3eb1b0_0
automat 20.2.0 py_0
autopep8 1.6.0 pyhd3eb1b0_1
babel 2.9.1 pyhd3eb1b0_0
backcall 0.2.0 pyhd3eb1b0_0
backports 1.1 pyhd3eb1b0_0
backports.functools_lru_cache 1.6.4 pyhd3eb1b0_0
backports.tempfile 1.0 pyhd3eb1b0_1
backports.weakref 1.0.post1 py_1
bcrypt 3.2.0 py39h2bbff1b_1
beautifulsoup4 4.11.1 py39haa95532_0
binaryornot 0.4.4 pyhd3eb1b0_1
bitarray 2.5.1 py39h2bbff1b_0
bkcharts 0.2 py39haa95532_1
black 22.6.0 py39haa95532_0
blas 1.0 mkl
bleach 4.1.0 pyhd3eb1b0_0
blosc 1.21.0 h19a0ad4_1
bokeh 2.4.3 py39haa95532_0
boto3 1.24.28 py39haa95532_0
botocore 1.27.28 py39haa95532_0
bottleneck 1.3.5 py39h080aedc_0
brotli 1.0.9 h2bbff1b_7
brotli-bin 1.0.9 h2bbff1b_7
brotlipy 0.7.0 py39h2bbff1b_1003
bzip2 1.0.8 he774522_0
ca-certificates 2022.07.19 haa95532_0
certifi 2022.9.14 py39haa95532_0
cffi 1.15.1 py39h2bbff1b_0
cfitsio 3.470 h2bbff1b_7
chardet 4.0.0 py39haa95532_1003
charls 2.2.0 h6c2663c_0
charset-normalizer 2.0.4 pyhd3eb1b0_0
click 8.0.4 py39haa95532_0
cloudpickle 2.0.0 pyhd3eb1b0_0
clyent 1.2.2 py39haa95532_1
colorama 0.4.5 py39haa95532_0
colorcet 3.0.0 py39haa95532_0
comtypes 1.1.10 py39haa95532_1002
conda 23.1.0 py39haa95532_0
conda-build 3.22.0 py39haa95532_0
conda-content-trust 0.1.3 py39haa95532_0
conda-env 2.6.0 haa95532_1
conda-pack 0.6.0 pyhd3eb1b0_0
conda-package-handling 2.0.2 py39haa95532_0
conda-package-streaming 0.7.0 py39haa95532_0
conda-repo-cli 1.0.27 py39haa95532_0
conda-token 0.4.0 pyhd3eb1b0_0
conda-verify 3.4.2 py_1
console_shortcut 0.1.1 4
constantly 15.1.0 pyh2b92418_0
cookiecutter 1.7.3 pyhd3eb1b0_0
cryptography 37.0.1 py39h21b164f_0
cssselect 1.1.0 pyhd3eb1b0_0
cuda-cccl 12.1.55 0 nvidia
cuda-cudart 11.8.89 0 nvidia
cuda-cudart-dev 11.8.89 0 nvidia
cuda-cupti 11.8.87 0 nvidia
cuda-libraries 11.8.0 0 nvidia
cuda-libraries-dev 11.8.0 0 nvidia
cuda-nvrtc 11.8.89 0 nvidia
cuda-nvrtc-dev 11.8.89 0 nvidia
cuda-nvtx 11.8.86 0 nvidia
cuda-profiler-api 12.1.55 0 nvidia
cuda-runtime 11.8.0 0 nvidia
cudatoolkit 10.1.243 h74a9793_0
curl 7.84.0 h2bbff1b_0
cycler 0.11.0 pyhd3eb1b0_0
cython 0.29.32 py39hd77b12b_0
cytoolz 0.11.0 py39h2bbff1b_0
daal4py 2021.6.0 py39h757b272_1
dal 2021.6.0 h59b6b97_874
dask 2022.7.0 py39haa95532_0
dask-core 2022.7.0 py39haa95532_0
dataclasses 0.8 pyh6d0b6a4_7
datashader 0.14.1 py39haa95532_0
datashape 0.5.4 py39haa95532_1
debugpy 1.5.1 py39hd77b12b_0
decorator 5.1.1 pyhd3eb1b0_0
defusedxml 0.7.1 pyhd3eb1b0_0
diff-match-patch 20200713 pyhd3eb1b0_0
dill 0.3.4 pyhd3eb1b0_0
distributed 2022.7.0 py39haa95532_0
docutils 0.18.1 py39haa95532_3
entrypoints 0.4 py39haa95532_0
et_xmlfile 1.1.0 py39haa95532_0
ffmpeg 1.4 pypi_0 pypi
ffmpeg-python 0.2.0 pypi_0 pypi
fftw 3.3.9 h2bbff1b_1
filelock 3.6.0 pyhd3eb1b0_0
flake8 4.0.1 pyhd3eb1b0_1
flask 1.1.2 pyhd3eb1b0_0
fonttools 4.25.0 pyhd3eb1b0_0
freetype 2.10.4 hd328e21_0
fsspec 2022.7.1 py39haa95532_0
future 0.18.2 py39haa95532_1
gensim 4.1.2 py39hd77b12b_0
giflib 5.2.1 h62dcd97_0
glob2 0.7 pyhd3eb1b0_0
greenlet 1.1.1 py39hd77b12b_0
h5py 3.7.0 py39h3de5c98_0
hdf5 1.10.6 h1756f20_1
heapdict 1.0.1 pyhd3eb1b0_0
holoviews 1.15.0 py39haa95532_0
hvplot 0.8.0 py39haa95532_0
hyperlink 21.0.0 pyhd3eb1b0_0
icc_rt 2022.1.0 h6049295_2
icu 58.2 ha925a31_3
idna 3.3 pyhd3eb1b0_0
imagecodecs 2021.8.26 py39hc0a7faf_1
imageio 2.19.3 py39haa95532_0
imagesize 1.4.1 py39haa95532_0
importlib-metadata 4.11.3 py39haa95532_0
importlib_metadata 4.11.3 hd3eb1b0_0
incremental 21.3.0 pyhd3eb1b0_0
inflection 0.5.1 py39haa95532_0
iniconfig 1.1.1 pyhd3eb1b0_0
intake 0.6.5 pyhd3eb1b0_0
intel-openmp 2021.4.0 haa95532_3556
intervaltree 3.1.0 pyhd3eb1b0_0
ipykernel 6.15.2 py39haa95532_0
ipython 7.31.1 py39haa95532_1
ipython_genutils 0.2.0 pyhd3eb1b0_1
ipywidgets 7.6.5 pyhd3eb1b0_1
isort 5.9.3 pyhd3eb1b0_0
itemadapter 0.3.0 pyhd3eb1b0_0
itemloaders 1.0.4 pyhd3eb1b0_1
itsdangerous 2.0.1 pyhd3eb1b0_0
jdcal 1.4.1 pyhd3eb1b0_0
jedi 0.18.1 py39haa95532_1
jellyfish 0.9.0 py39h2bbff1b_0
jinja2 2.11.3 pyhd3eb1b0_0
jinja2-time 0.2.0 pyhd3eb1b0_3
jmespath 0.10.0 pyhd3eb1b0_0
joblib 1.1.0 pyhd3eb1b0_0
jpeg 9e h2bbff1b_0
jq 1.6 haa95532_1
json5 0.9.6 pyhd3eb1b0_0
jsonschema 4.16.0 py39haa95532_0
jupyter 1.0.0 py39haa95532_8
jupyter_client 7.3.4 py39haa95532_0
jupyter_console 6.4.3 pyhd3eb1b0_0
jupyter_core 4.11.1 py39haa95532_0
jupyter_server 1.18.1 py39haa95532_0
jupyterlab 3.4.4 py39haa95532_0
jupyterlab_pygments 0.1.2 py_0
jupyterlab_server 2.10.3 pyhd3eb1b0_1
jupyterlab_widgets 1.0.0 pyhd3eb1b0_1
keyring 23.4.0 py39haa95532_0
kiwisolver 1.4.2 py39hd77b12b_0
lazy-object-proxy 1.6.0 py39h2bbff1b_0
lcms2 2.12 h83e58a3_0
lerc 3.0 hd77b12b_0
libaec 1.0.4 h33f27b4_1
libarchive 3.6.1 hebabd0d_0
libbrotlicommon 1.0.9 h2bbff1b_7
libbrotlidec 1.0.9 h2bbff1b_7
libbrotlienc 1.0.9 h2bbff1b_7
libcublas 11.11.3.6 0 nvidia
libcublas-dev 11.11.3.6 0 nvidia
libcufft 10.9.0.58 0 nvidia
libcufft-dev 10.9.0.58 0 nvidia
libcurand 10.3.2.56 0 nvidia
libcurand-dev 10.3.2.56 0 nvidia
libcurl 7.84.0 h86230a5_0
libcusolver 11.4.1.48 0 nvidia
libcusolver-dev 11.4.1.48 0 nvidia
libcusparse 11.7.5.86 0 nvidia
libcusparse-dev 11.7.5.86 0 nvidia
libdeflate 1.8 h2bbff1b_5
libiconv 1.16 h2bbff1b_2
liblief 0.11.5 hd77b12b_1
libnpp 11.8.0.86 0 nvidia
libnpp-dev 11.8.0.86 0 nvidia
libnvjpeg 11.9.0.86 0 nvidia
libnvjpeg-dev 11.9.0.86 0 nvidia
libpng 1.6.37 h2a8f88b_0
libsodium 1.0.18 h62dcd97_0
libspatialindex 1.9.3 h6c2663c_0
libssh2 1.10.0 hcd4344a_0
libtiff 4.4.0 h8a3f274_0
libuv 1.44.2 h2bbff1b_0
libwebp 1.2.2 h2bbff1b_0
libxml2 2.9.14 h0ad7f3c_0
libxslt 1.1.35 h2bbff1b_0
libzopfli 1.0.3 ha925a31_0
llvmlite 0.38.0 py39h23ce68f_0
locket 1.0.0 py39haa95532_0
lxml 4.9.1 py39h1985fb9_0
lz4 3.1.3 py39h2bbff1b_0
lz4-c 1.9.3 h2bbff1b_1
lzo 2.10 he774522_2
m2-msys2-runtime 2.5.0.17080.65c939c 3
m2-patch 2.7.5 2
m2w64-libwinpthread-git 5.0.0.4634.697f757 2
markdown 3.3.4 py39haa95532_0
markupsafe 2.0.1 py39h2bbff1b_0
matplotlib 3.5.2 py39haa95532_0
matplotlib-base 3.5.2 py39hd77b12b_0
matplotlib-inline 0.1.6 py39haa95532_0
mccabe 0.6.1 py39haa95532_2
menuinst 1.4.19 py39h59b6b97_0
mistune 0.8.4 py39h2bbff1b_1000
mkl 2021.4.0 haa95532_640
mkl-service 2.4.0 py39h2bbff1b_0
mkl_fft 1.3.1 py39h277e83a_0
mkl_random 1.2.2 py39hf11a4ad_0
mock 4.0.3 pyhd3eb1b0_0
more-itertools 9.1.0 pypi_0 pypi
mpmath 1.2.1 py39haa95532_0
msgpack-python 1.0.3 py39h59b6b97_0
msys2-conda-epoch 20160418 1
multipledispatch 0.6.0 py39haa95532_0
munkres 1.1.4 py_0
mypy_extensions 0.4.3 py39haa95532_1
navigator-updater 0.3.0 py39haa95532_0
nbclassic 0.3.5 pyhd3eb1b0_0
nbclient 0.5.13 py39haa95532_0
nbconvert 6.4.4 py39haa95532_0
nbformat 5.5.0 py39haa95532_0
nest-asyncio 1.5.5 py39haa95532_0
networkx 2.8.4 py39haa95532_0
nltk 3.7 pyhd3eb1b0_0
nose 1.3.7 pyhd3eb1b0_1008
notebook 6.4.12 py39haa95532_0
numba 0.55.1 py39hf11a4ad_0
numexpr 2.8.3 py39hb80d3ca_0
numpy 1.21.5 py39h7a0a035_3
numpy-base 1.21.5 py39hca35cd5_3
numpydoc 1.4.0 py39haa95532_0
olefile 0.46 pyhd3eb1b0_0
openai-whisper 20230314 pypi_0 pypi
openjpeg 2.4.0 h4fc8c34_0
openpyxl 3.0.10 py39h2bbff1b_0
openssl 1.1.1q h2bbff1b_0
packaging 21.3 pyhd3eb1b0_0
pandas 1.4.4 py39hd77b12b_0
pandocfilters 1.5.0 pyhd3eb1b0_0
panel 0.13.1 py39haa95532_0
param 1.12.0 pyhd3eb1b0_0
paramiko 2.8.1 pyhd3eb1b0_0
parsel 1.6.0 py39haa95532_0
parso 0.8.3 pyhd3eb1b0_0
partd 1.2.0 pyhd3eb1b0_1
pathlib 1.0.1 pyhd3eb1b0_1
pathspec 0.9.0 py39haa95532_0
patsy 0.5.2 py39haa95532_1
pep8 1.7.1 py39haa95532_1
pexpect 4.8.0 pyhd3eb1b0_3
pickleshare 0.7.5 pyhd3eb1b0_1003
pillow 9.2.0 py39hdc2b20a_1
pip 22.2.2 py39haa95532_0
pkginfo 1.8.2 pyhd3eb1b0_0
platformdirs 2.5.2 py39haa95532_0
plotly 5.9.0 py39haa95532_0
pluggy 1.0.0 py39haa95532_1
powershell_shortcut 0.0.1 3
poyo 0.5.0 pyhd3eb1b0_0
prometheus_client 0.14.1 py39haa95532_0
prompt-toolkit 3.0.20 pyhd3eb1b0_0
prompt_toolkit 3.0.20 hd3eb1b0_0
protego 0.1.16 py_0
psutil 5.9.0 py39h2bbff1b_0
ptyprocess 0.7.0 pyhd3eb1b0_2
py 1.11.0 pyhd3eb1b0_0
py-lief 0.11.5 py39hd77b12b_1
pyasn1 0.4.8 pyhd3eb1b0_0
pyasn1-modules 0.2.8 py_0
pycodestyle 2.8.0 pyhd3eb1b0_0
pycosat 0.6.3 py39h2bbff1b_0
pycparser 2.21 pyhd3eb1b0_0
pyct 0.4.8 py39haa95532_1
pycurl 7.45.1 py39hcd4344a_0
pydispatcher 2.0.5 py39haa95532_2
pydocstyle 6.1.1 pyhd3eb1b0_0
pyerfa 2.0.0 py39h2bbff1b_0
pyflakes 2.4.0 pyhd3eb1b0_0
pygments 2.11.2 pyhd3eb1b0_0
pyhamcrest 2.0.2 pyhd3eb1b0_2
pyjwt 2.4.0 py39haa95532_0
pylint 2.14.5 py39haa95532_0
pyls-spyder 0.4.0 pyhd3eb1b0_0
pynacl 1.5.0 py39h8cc25b3_0
pyodbc 4.0.34 py39hd77b12b_0
pyopenssl 22.0.0 pyhd3eb1b0_0
pyparsing 3.0.9 py39haa95532_0
pyqt 5.9.2 py39hd77b12b_6
pyrsistent 0.18.0 py39h196d8e1_0
pysocks 1.7.1 py39haa95532_0
pytables 3.6.1 py39h56d22b6_1
pytest 7.1.2 py39haa95532_0
python 3.9.13 h6244533_1
python-dateutil 2.8.2 pyhd3eb1b0_0
python-fastjsonschema 2.16.2 py39haa95532_0
python-libarchive-c 2.9 pyhd3eb1b0_1
python-lsp-black 1.0.0 pyhd3eb1b0_0
python-lsp-json
小智 8
您需要确保安装的PyTorch 支持 CUDA，才能真正利用您的 GPU。
请参阅此处的参考文档：https ://pytorch.org/get-started/locally/#with-cuda-1
python3 -m pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
python3 -m pip3 install -U openai-whisper
然后您可以在 Python 中确认 CUDA 是否可用
import torch
torch.cuda.is_available()