Deploying audio classification model on computer or mobile phone not working?

Report bugs
#1

i have succesfully created a audio classification model by following this tutorial: Build Your Own ML-Powered Keyword Spotting Model in 30KB RAM - YouTube
. Now when i go to deployment > run impulse on computer > click build. A new tab opens > i give acces to microphone > its starts listening for 0.5 seconds and after that an error pops up:
Failed to load
Classification failed (err code: -5)

In the console on the browser i get the following:
ERR: Failed to run DSP process (-1013)

Now when i switch to data collection mode, it works just fine. But the live classification mode doesen’t work. The same error occurs when i run impulse on mobile phone and also on raspberry pi 4 with the edge-impulse-linux-runner command.

project id: 198363

#2

Hello @ziekejuan344,

Can you tell me which DSP block you are using?

The error details can be found here.

If you’re using MFCC, try using MFE instead is you’re using audio. In the meantime, our core engineering team will have a look.

Let me know if that solve your issue, other people had the same issue.

Best,

Louis

Some issues with Arduino Nano 33 BLE Sense
Arduino Nano 33 BLE Sense Rev2
#3

It did not work with MFCC. I changed it to MFE and it works now.

1 Like
#4

I have the same problem with an audio detection model (public):

I am already using MFE. Everything works until Deployment, but then desktop or mobile give the error: Classification failed (err code: -5)

I can download the .eim and and run in the terminal:

% python classify.py mubarik-project-1-mac-x86_64-v18.eim
Loaded runner for "Cumhur Erkut / Mubarik-project-1"
0 --> Red Microphone
1 --> Earmuffs
3 --> MacBook Pro Microphone
5 --> Microsoft Teams Audio
6 --> Descript Loopback Recorder
7 --> ZoomAudioDevice
Type the id of the audio device you want to use: 
1
selected Audio device: 1
Result (8 ms.) cardboard: 0.00	glass: 0.00	metal: 0.00	noise: 1.00	
Result (2 ms.) cardboard: 0.00	glass: 0.00	metal: 0.00	noise: 1.00

I can also download & run the WebAssambly on a browser with feature pasting. But the desktop / mobile deployments fail.

Could you please check what is causing this @louis ? Thanks.