Abstract: A method and an apparatus are disclosed for a speaker verification. The apparatus may comprise at least one processor configured to execute instructions to perform generating, based on an utterance of a speaker to a deep learning-based speaker verification model, a speaker embedding with a preset dimension, verifying, based on the speaker embedding, the speaker, and detecting, based on the verified speaker, an identity of a user associated with the apparatus.

Abstract: A method and an apparatus for separating speeches based on local and global modeling network (LAGNet) are provided. A speech separation apparatus includes an encoder configured to progressively compress a sequence of mixed speeches by using a one-dimensional convolution-based local block to generate local information, a bottleneck configured to generate global information by using a multiple self attention-based global block, and a decoder configured to progressively reconstruct the sequence by using the local block. A speech separation apparatus also utilizes a skip connection that includes gates configured to filter the local information by using the global information.

Abstract: A photomask for a photolithography process includes a mask substrate, a reflective multilayer on the mask substrate, and a light absorber pattern on the reflective multilayer and having hole patterns, wherein the hole patterns include a main hole pattern for pattern transfer onto a wafer, first sub-resolution assist feature (SRAF) hole patterns arranged at regular intervals to provide honeycomb lattices in a first region centered around the main hole pattern and having a first pitch less than or equal to a diffraction limit in the photolithography process, and second SRAF hole patterns arranged at regular intervals to surround the main hole pattern and the first SRAF patterns and providing honeycomb lattices in a second region centered around the main hole pattern and surrounding the first region, the second SRAF hole patterns being arranged with a second pitch less than or equal to the diffraction limit in the photolithography process.

Abstract: A method of a vehicle may comprise: based on presence of a first user and a second user in the vehicle, determining that the first user is a driver and the second user is a passenger; applying, based on a first profile of the first user and based on the first user being the driver, profile settings of the first profile to the vehicle; while applying the profile settings of the first profile to the vehicle, detecting a voice command of the second user; and while maintaining at least one profile setting of the profile settings of the first profile to the vehicle: determining, from profile settings of a second profile of the second user and based on the voice command, data for executing the voice command; and executing, based on the data, at least one operation associated with the voice command.

Abstract: A method and a device perform user authentication of an occupant in a vehicle based on speech recognition. The device includes a memory that stores a reference embedding set including a feature embedding for a registered user's utterance and feature embeddings for synthesis results between the registered user's utterance and a plurality of environmental noises. The device also includes a user interface that receives input audio including utterance of an occupant of the vehicle and noise, and a processor that transforms the input audio to an input embedding and determines whether the occupant is a registered user based on a comparison between the input embedding and the reference embedding set.

Abstract: Provided is a sound source localization method including steps of: (a) receiving a mixed signal of a target sound source signal and noise and echo signals through multiple microphones including at least two microphones; (b) generating a binarized mask based on a diffuseness by using a coherence-to-diffuseness ratio CDR, which is information on the target sound source and the noise source, by using the input signal; (c) pre-processing an input signal to multiple microphones by using the generated binarized mask; and (d) performing a predetermined algorithm such as the GCC-PHAT or the SRP-PHAT on the pre-processed input signal to estimate a direction of the target sound source.

Abstract: Provided is a sound source localization method including steps of: (a) receiving a mixed signal of a target sound source signal and noise and echo signals through multiple microphones including at least two microphones; (b) generating a binarized mask based on a diffuseness by using a coherence-to-diffuseness ratio CDR, which is information on the target sound source and the noise source, by using the input signal; (c) pre-processing an input signal to multiple microphones by using the generated binarized mask; and (d) performing a predetermined algorithm such as the GCC-PHAT or the SRP-PHAT on the pre-processed input signal to estimate a direction of the target sound source.