AI Sleep Sound Machine powered by Jieli Audio DSP

Client

A wellness brand creating a sleep aid device that dynamically adjusts white noise based on the user's sleep cycles.

Challenge

Detecting sleep phases using contactless acoustic analysis (snoring/breathing detection) and altering audio frequencies in real-time.

Product Concept & Architecture

We architected the system around the Jieli AC695N Audio DSP. It provides enough computational headroom to run a basic acoustic classifier while driving a high-fidelity 5W speaker.

Engineering Solution

The device uses an onboard microphone to monitor room acoustics. We trained a lightweight SVM (Support Vector Machine) classifier to run on the AC695N, categorizing sounds into 'quiet breathing', 'snoring', or 'environmental disturbance'. If a disturbance is detected, the DSP automatically shifts the equalizer and volume of the white noise to mask the intrusion, using smooth volume ramping to avoid waking the user.

Result

Non-wearable sleep tracking and acoustic masking working completely offline. Proven to reduce sleep disruptions by 40% in clinical trials.