openVINO transcription fails on NPU

Just started playing with OpenVINO models. Was looking to try out the transcribe/translate features specifically. I can get the model to transcribe on CPU but whenever I select NPU it gives an error. Log details below. Any help on what I am doing and how to fix will be helpful.

06:59:08: Audacity 3.4.2
Log Cleared.
06:59:29: Operation ‘Pre-processing’ took 0.000000 seconds. Poll was called 0 times and took 0.000000 seconds. Yield was called 0 times and took 0.000000 seconds.
06:59:32: Error: In Whisper Transcription Effect, exception: whisper.cpp context creation / initialization failed
06:59:33: Operation ‘OpenVINO Whisper Transcription’ took 3.128000 seconds. Poll was called 701 times and took 0.131200 seconds. Yield was called 37 times and took 0.076664 seconds.
06:59:32: Error: whisper_ctx_init_openvino_encoder failed for device = NPU

Do you mean GPU?

Hi @AIdemoexplorer,

Is there any more to the log, or is that it?

Can you let me know which NPU driver version you are using? You can find this in the task manager, if you click ‘Performance’ on the left, and then click on ‘NPU’. There should be a ‘Driver Version’ listed toward the bottom of the window.


Thanks for offering to help. I re-ran to confirm whether log shows any other details. It’s same as what I shared previously.

The NPU driver version is

I am running Intel’s recently announced Meteor Lake processor which has an integrated NPU.

CPU and GPU work just fine. It’s the NPU where I am running in to an error.

For me, CPU works just fine, but the GPU parameter takes forever!

For NPU support with these OpenVINO plugins, you’ll need driver version…

Here’s the link to download the driver.

Let me know if this resolves your issue. Thanks!

This did fix the issue. Thanks a bunch!

1 Like

This topic was automatically closed after 29 days. New replies are no longer allowed.