US 12,149,890 B2
Neural network-driven frequency translation
Kelly Fitz, Eden Prairie, MN (US); Tao Zhang, Eden Prairie, MN (US); Buye Xu, Minnetonka, MN (US); and Mohammad Abdollahi, Tbilisi (GE)
Assigned to Starkey Laboratories, Inc., Eden Prairie, MN (US)
Filed by Starkey Laboratories, Inc., Eden Prairie, MN (US)
Filed on Jul. 6, 2023, as Appl. No. 18/347,979.
Application 18/347,979 is a continuation of application No. 17/647,532, filed on Jan. 10, 2022, granted, now 11,736,870.
Application 17/647,532 is a continuation of application No. 16/797,157, filed on Feb. 21, 2020, granted, now 11,223,909.
Application 16/797,157 is a continuation of application No. 15/092,487, filed on Apr. 6, 2016, granted, now 10,575,103.
Claims priority of provisional application 62/146,202, filed on Apr. 10, 2015.
Prior Publication US 2024/0007800 A1, Jan. 4, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. H04R 25/00 (2006.01); G06N 3/08 (2023.01); G10L 25/84 (2013.01)
CPC H04R 25/353 (2013.01) [G06N 3/08 (2013.01); G10L 25/84 (2013.01); H04R 25/507 (2013.01); H04R 25/55 (2013.01); H04R 2225/021 (2013.01); H04R 2225/023 (2013.01); H04R 2225/025 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A hearing device comprising:
a microphone configured to receive an input signal and provide an audio signal based on the input signal; and
a processor configured to:
process the audio signal to correct for a hearing impairment of a wearer; and
control frequency translation of the audio signal using a neural network, the neural network trained, on a device external to the hearing device, to identify acoustic features in a plurality of audio signals and predict outputs for the plurality of audio signals by mapping features of the plurality of audio signals to a set of algorithm parameters for producing frequency-lowered speech cues.