公开(公告)号：US20230419114A1

公开(公告)日：2023-12-28

申请号：US18466469

申请日：2023-09-13

Applicant: Samsung Electronics Co., Ltd.

Inventor： Krzysztof ARENDT ,  Grzegorz STEFANSKI ,  Piotr MASZTALSKI ,  Artur SZUMACZUK ,  Jakub TKACZUK ,  Mateusz MATUSZEWSKI ,  Michal SWIATEK ,  Tymoteusz OLENIECKI

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: An electronic apparatus is provided. The electronic apparatus includes a memory and a processor, wherein the processor is configured to, by executing the at least one instruction, acquire a plurality of training data; acquire a plurality of embedding vectors that are mappable to an embedding space for the plurality of training data, respectively; train an artificial intelligence model classifying the plurality of training data based on the plurality of embedding vectors, identify an embedding vector misclassified by the artificial intelligence model among the plurality of embedding vectors, identify an embedding vector closest to the misclassified embedding vector in the embedding space, acquire a synthetic embedding vector corresponding to a path connecting the misclassified embedding vector to the embedding vector closest to the misclassified embedding vector in the embedding space, and re-train the artificial intelligence model by adding the synthetic embedding vector to the training data.

公开(公告)号：US20250029623A1

公开(公告)日：2025-01-23

申请号：US18745390

申请日：2024-06-17

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Tomasz KRUKOWICZ ,  Marcin GOZDZIEWSKI ,  Krzysztof ARENDT ,  Jakub TKACZUK

IPC: G10L21/0208

Abstract: An example electronic apparatus includes a memory configured to store at least one instruction and at least one processor connected to the memory to control the electronic apparatus. The at least one processor is configured to, by executing the at least one instruction, obtain a first audio signal including a voice signal and a noise signal, convert the first audio signal in a time domain to a second audio signal in a frequency domain, obtain a first gain value representing a Signal-to-Noise Ratio (SNR) from the second audio signal, obtain a second gain value with a first dynamic range by filtering the first gain value, obtain a third gain value by inputting the second gain value to a neural network model trained to output a signal from which noise is removed, and convert the second audio signal to a third audio signal from which at least a portion of the noise signal is removed, using the third gain value.