I just tried out the new EON Tuner (which is really cool btw!), and got the following error when trying to record and load samples for the live classification:
The model that I chose to use from the EON Tuner is called “mfe-conv1d-2c2”. This model uses an Audio (MFE) processing block and a Neural Network (Keras) learning block.
I’ve tried recording with my mobile phone, the nRF52840DK and the nRF5340DK, with the IKS02A1 shield connected when recording with the nRF DKs. The sampling frequency has been set to 16 kHz for all devices, which is the same as the sampling frequency of the uploaded datasets.
Any idea of how to fix this issue?
Thanks for reporting this. A fix for this issue will be deployed tomorrow.
As a temporary work-around try to ‘Select’ a different model variant as primary model on the EON tuner page, and switch back to the current model variant afterwards. This will most likely resolve the issue. If it does not, please let me know and i’ll try to help further.
Thank you for the tip, @mathijs!
I tried what you suggested, and I also tried to us another model when doing the live classification. Unfortunately neither of these fixes seemed to work for me. I still got the same error when trying to load a sample. Both when I tried with a sample recorded earlier (when I first got the error) and a sample recorded just now.
I fixed this issue in the database for your project and verified ‘live classification’ works normally now in your project. A code-fix has also deployed now to prevent this issue from re-occurring.
Please let me know if there is anything else I can help with.