Citric / xiao-esp32s3-speech Public
Target: Cortex-M4F 80MHz Clone this project

Test data

Set confidence thresholds
Model variants
Test 'Unoptimized (float32)' model Test 'Quantized (int8)' model
Set the 'expected outcome' for each sample to the desired outcome to automatically score the impulse.
Sample name Expected outcome Length Accuracy Result
stop.6 stop 10s 26% 14 noise, 5 stop
Show classification Show in data acquisition
stop.4 stop 10s 21% 14 noise, 4 stop, 1 uncertain
Show classification Show in data acquisition
stop.10 stop 10s 21% 14 noise, 4 stop, 1 uncertain
Show classification Show in data acquisition
noise.16 noise 10s 100% 19 noise
Show classification Show in data acquisition
noise.12 noise 10s 100% 19 noise
Show classification Show in data acquisition
hello.9 hello 10s 21% 15 noise, 4 hello
Show classification Show in data acquisition
hello.7 hello 10s 11% 16 noise, 2 hello, 1 uncertain
Show classification Show in data acquisition
hello.10 hello 10s 26% 14 noise, 5 hello
Show classification Show in data acquisition
stop.1.s5 stop 1s 0% 1 uncertain
Show classification Show in data acquisition
stop.3.s5 stop 1s 100% 1 stop
Show classification Show in data acquisition
stop.7.s2 stop 1s 100% 1 stop
Show classification Show in data acquisition
stop.8.s3 stop 1s 100% 1 stop
Show classification Show in data acquisition
noise.7 noise 10s 95% 18 noise, 1 stop
Show classification Show in data acquisition
noise.4 noise 10s 100% 19 noise
Show classification Show in data acquisition
noise.2 noise 10s 100% 19 noise
Show classification Show in data acquisition
noise.6 noise 10s 100% 19 noise
Show classification Show in data acquisition
noise.1 noise 10s 100% 19 noise
Show classification Show in data acquisition

Model testing output

Results

Model version:
Accuracy
63.35%

Confusion matrix

Loading...

Feature explorer

Loading...
Help & documentation Get help on the forum Send feedback Contact sales
Join us
Terms of Service Privacy policy