Feature Extraction
Multi-channel auditory feature extraction
AUDITORY AI SOC PLATFORMCHANNELS2,001 UPDATE INTERVAL2ms FREQUENCY RESOLUTION~2Hz-class FSE AFC WDRC RNN-SE
From sound to the features that matter.
Connecting feature extraction and auditory signal-processing functions in a common core.
FEATURE EXTRACTION
Three parameters for auditory analysis.
Feature Update
Feature-update interval
Frequency Analysis
Frequency-analysis resolution
Development specifications. The 2ms figure is a feature-update interval, not end-to-end system latency.
AUDITORY PROCESSING
Processing functions for listening environments.
Frontal speech enhancement
Uses the direction of incoming sound to focus processing on frontal speech.
Acoustic feedback control
Processes sound returning from the speaker to the microphone.
Wide dynamic range compression
Adjusts gain according to the input sound level.
Neural-network speech enhancement
Speech-enhancement technology using a recurrent neural network is in development.
DISCUSS YOUR APPLICATION
Discuss technology and developmentDesign how your next device hears.
Tell us about your device and its auditory-processing requirements.