Projects - Jinkyoung Hwangbo

Audio Eraser for Galaxy S25

Mar 2024 - Present
Samsung Research. Commercialized

I developed a lightweight, on-device sound detection model for Samsung’s Audio Eraser feature on the Galaxy S25 series.

Anomaly Sound Detection for Home Appliances

Oct 2022 - Dec 2024
Samsung Research. Commercialized

An AI-powered diagnostic system of home appliances, released in the domestic market.

Comparison of Recognizing Emotions Depending on Bandwidth-limitation of Digital Speech

Mar - Jun 2022
Bachelor's Thesis in Industrial Engineering, Seoul National University.

An empirical study examining whether bandwidth limitations in digital speech affect emotional transmission.

[Paper (KOR)]

Smootify: Infinite AI Music Streaming

Jan - Jun 2021
Tobigs, Big Data Study Group

Smootify is an AI-powered music streaming project that generates seamless, natural transitions between songs using MIDI data and deep learning, enabling infinite and uninterrupted listening experiences.

[code] [demo]

GAN based Sign Language Synthesis Model

May - Dec 2021
Undergraduate Research Program, Seoul National University.

A GAN-based framework that synthesizes sign language videos by transferring interpreter motions to speakers while preserving identity and facial expressions, enhancing accessibility for deaf audiences.

[paper (KOR)]

MPTI: Meet Persona Through AI

Jun - Dec 2021
Tobigs, Big Data Study Group

MPTI is a conversational AI project that lets users chat with movie character personas, like Sherlock Holmes and Iron Man, using chatbots fine-tuned on their movie dialogues for realistic, character-driven interactions.

[code]

Speech Bandwidth Extension with Hifi-GAN

Jul - Dec 2021
Bachelor's Thesis in Artificial Intelligence, Samsung Research.

A GAN-based deep learning model for speech bandwidth extension that converts wideband audio to super-wideband, achieving 30% MUSHRA score improvement with robust speaker generalization.

[Paper (KOR)]