Overview of Algorithm & Data
Chop sound signals into frames:
5,000 signals per second
500 signals per frame
600 frames per mins
Overview of Algorithm:
Derive per min sleep/wake states based on received sound signals.
Training Data : temperature, light, sound, time -> state
5,000 signals per second
500 signals per frame
600 frames per mins
Overview of Algorithm:
Derive per min sleep/wake states based on received sound signals.
Training Data : temperature, light, sound, time -> state
Sound Signals -> Labels of Training Data
Body Movement Data Distribution (left) v.s. Other Data (eg. car passing, snoring) Distribution (right)
State Determination:
D >= 1, then "wake", otherwise, "sleep"
Sleep Quality per Night
Different Environments Adaptability
(rlh and var are the same)
mean(rms), std(rms) are calculated from noise frames of the current environment.
The actual used features of events are normalized by current environment noise.
So better adaptability for different environments.
mean(rms), std(rms) are calculated from noise frames of the current environment.
The actual used features of events are normalized by current environment noise.
So better adaptability for different environments.