Skip to content

Using Whisper OpenAI transcription only available offline even when installed? #7192

@JalalAl-Tamimi

Description

@JalalAl-Tamimi

Hello,
I have been trying to use the demo "microphone" with the automatic transcription using Whisper OpenAI, however, when I launch the experiment, I get an error message as:
Alert 4610:
Audio transcription service "Whisper" is not supported offline.
For further info see https://psychopy.org/alerts/4610.html
Alert 4610:
Audio transcription service "Whisper" is not supported offline.
For further info see https://psychopy.org/alerts/4610.html

I checked that psychopy-whisper is already installed. I uninstalled it and reinstalled it by using the plugin/package manager, then packages to uninstall.
I checked the instructions for installing whisper for offline usage (see here: openai/whisper#1463) and went through "Step 2B: Whisper Install (Online Install for later Offline Use)". To do that, I used the "Open PIP terminal" from within psychopy and reinstalled whisper. I also went through Step 3: Download Other Required Files (for Offline Use), which consists of downloading the various models (I have the tiny and the base).
I made sure to change the settings of the microphone to use either the tiny or the base. I then regenerated the microphone.js script (just in case). I ran the experiment in pilot mode. I still get the Alert as above.
Then the experiment runs and gets to the first phrase, but then the window disappears with no error or warning.
I am using the latest version of PsychoPy (2025.1.0)

Let me know how to fix this?
Thanks
Jalal

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions