US 12,412,099 B2
Noise waveform removing device, model training device, noise waveform removing method, model training method, generation model, and wearable device
Masashi Ueda, Tokyo (JP)
Assigned to CASIO COMPUTER CO., LTD., Tokyo (JP)
Filed by CASIO COMPUTER CO., LTD., Tokyo (JP)
Filed on Sep. 14, 2021, as Appl. No. 17/474,981.
Claims priority of application No. 2020-157821 (JP), filed on Sep. 18, 2020.
Prior Publication US 2022/0092434 A1, Mar. 24, 2022
Int. Cl. G10K 11/16 (2006.01); G06N 3/045 (2023.01); G06N 3/088 (2023.01); G10K 15/02 (2006.01)
CPC G06N 3/088 (2013.01) [G06N 3/045 (2023.01); G10K 11/16 (2013.01); G10K 15/02 (2013.01)] 11 Claims
OG exemplary drawing
 
1. A noise waveform removing device comprising:
at least one processor, the at least one processor being configured to:
obtain noise-containing movement waveform data;
using a trained generation model, determine a filter contribution rate based on the noise-containing movement waveform data;
generate first intermediate movement waveform data by applying the filter contribution rate denoted as R to filtered noise-containing movement waveform data, the filtered noise-containing movement waveform data being obtained by applying low-pass filtering to the noise-containing movement waveform data;
generate second intermediate movement waveform data by applying (1−R) to the noise-containing movement waveform data; and
synthesize the first intermediate movement waveform data and the second intermediate movement waveform data, to generate noise-free movement waveform data.