Scriberr real Offline and On Premis

[stony007de]

HI

is correct that Scriberr is fully offline an Runs without any connection to the world-wide web?
If is deny the way out, no transcription is working anymore.

[stony007de]

if you check the Logs you will find

  available_backends = torchaudio.list_audio_backends()
   Building whisperx @ file:///app/whisperx-env/WhisperX
  × Failed to build `whisperx @ file:///app/whisperx-env/WhisperX`
  ├─▶ Failed to resolve requirements from `build-system.requires`
  ├─▶ No solution found when resolving: `setuptools`
  ├─▶ Failed to fetch: `https://pypi.org/simple/setuptools/`
  ├─▶ Request failed after 3 retries
  ├─▶ error sending request for url (https://pypi.org/simple/setuptools/)
  ╰─▶ operation timed out

so I need a connection?

[rishikanthc]

For whisper, the models aren't auto downloaded and cached during setup like how Nvidia models are handled.
Whisper models are downloaded the first time you initiate transcription. So for eg. Let's say after the app starts on a clean install, if you run a transcription job with whisper small, the small model gets downloaded and cached. On the next run the cached model will be used and you don't need to fetch the model again.