how to install language models - medium model in my case on Mac M1 - without getting "SHA256 checksum does not match" ?

[skywalk-explore]

The file upload up till 459 mb every time then I get the error.

I've seen other post about this error, but no solutions were identified. One person said they just restarted something and the issue resolved it self.

Any help ?

small.pt exists, but the SHA256 checksum does not match; re-downloading the file
warnings.warn(
99%|██████████████████████████████████████▊| 459M/461M [04:11<00:01, 1.91MiB/s]
Traceback (most recent call last):
File "/Users/parenti/PycharmProject/srtTranscribe/main.py", line 40, in
transcribe_and_srt("Video/Frieren - 5.mov")
File "/Users/parenti/PycharmProject/srtTranscribe/main.py", line 6, in transcribe_and_srt
model = whisper.load_model("small")
File "/Users/parenti/PycharmProject/srtTranscribe/venv/lib/python3.10/site-packages/whisper/init.py", line 133, in load_model
checkpoint_file = _download(_MODELS[name], download_root, in_memory)
File "/Users/parenti/PycharmProject/srtTranscribe/venv/lib/python3.10/site-packages/whisper/init.py", line 87, in _download
raise RuntimeError(
RuntimeError: Model has been downloaded but the SHA256 checksum does not not match. Please retry loading the model.

[glangford]

Try
cd ~/.cache/whisper

delete small.pt and try again

[glangford]

...if this still doesn't work, try running your test outside of pycharm or any other IDE. See also

• How to stop Whisper downloading the model each time #1013

[skywalk-explore]

Prior to deleting model the first line of the error is the warning, as below.

• 1. /Users/xx/xxx/xx: UserWarning: /Users/parenti/.cache/whisper/small.pt exists, but the SHA256 checksum does not match; re-downloading the file warnings.warn( 97%|██████████████████████████████████████▋ | 446M/461M [04:17<00:15, 982kiB/s])

After deleting the small model, the warning begins with:

• 50%|███████████████████▍ | 230M/461M [01:23<00:54, 4.45MiB/s]

then the rest of the error is identical to the original.

It doesn't seem to work. Even after deleting I get the same error but with the above difference.

Went to the link provided and there I could download and use the medium model, it works but too big for Mac. Take 70 something minutes to transcribe a couple of minutes to mp3.

[skywalk-explore]

Having looked at the checksums of my file and the required file. there is a difference in the checksum. for some reason I don't have the correct file to the one required by the package.

my small.pt file checksum: c39b7d8503417cf17df598db01964c50fff498b271d235f985d8bf90aab0c4ef

Checksum required by package: BzY8DmU6=0{>%Rpa?JkvJ6qF(V^F86#Xh7JUGMK}P<N0000

[skywalk-explore]

It's working now thanks man.

1. remove the small.pt from file using cd ~/.cache/whisper then rm small.pt

2. download the correct small.pt file from (

   whisper/whisper/__init__.py

   Line 17 in ba3f3cd

   _MODELS = {

   )

3. copy the download small.pt file into the whisper cache. cp /Users/xxx/Downloads/small.pt ~/.cache/whisper/

Thanks again bro.