top of page

Bixby - Samsung's Intelligent Assistant

Elevating Voice Interaction with Localization and Intelligent Engineering

Introduction to the Project (2018 - 2019)

Overview:

The Samsung project focused on localization in Germany and enhancing the capabilities of their voice assistant, Bixby. As an NLP engineer and voice UX designer, my role was pivotal in ensuring the seamless user interactions with Bixby.

My Responsibilities as an NLU Engineer:

  • Data Collection and Processing: Gathered and processed large datasets to improve the voice assistant's understanding and responses.

  • Capsule NLU Training: Developed and trained capsules to handle various user intents and improve interaction accuracy.

  • Programming: Utilized Python and Java to create and optimize internal tools.

    • Internal Response Management Tool: Worked on this tool to streamline handling Bixby's responses, making it easier for language experts to manage.

    • Scripts for Workflow Improvement: Programmed scripts in Python to enhance internal workflows and processes, including generating scripts and conversion files from Bixby files into CSVs and vice versa.

    • Analyzing Bixby Capsules: Created scripts for analyzing NLU training capsules to improve their performance.

    • Internal Slack chatbot: Developed an internal slide board to facilitate project management and tracking.

My responsibilities as a Voice UX Designer:

  • Localization and Creation: Provided user experience design for local content providers, localizing German capsules and creating new ones to ensure a good voice user experience.

  • Voice Flows: Created voice flows for German capsules, including prompts and NLU utterances.

  • Usability and Cultural Adaptation: Ensured the usability of the capsules met the expectations of the German persona and cultural differences, adapting the Korean UX to fit the German market perfectly.

Key Use Cases:

Due to NDAs, I cannot provide specific details. 

  • IFA Capsule: Developed for the IFA in Berlin in 2018.

  • Bike Sharing App: Enabled voice control for an app designed for bike sharing.

  • News App: Implemented voice control for a news application.

  • General Use Cases: Enhanced various applications with voice interaction capabilities.

Outcome:

The project resulted in a highly effective and localized German version of Bixby. This involved rigorous data analysis, capsule training, and UX design to deliver a seamless voice user experience that meets German cultural expectations.

Introduction to the Project

Screenshot 2024-07-03 114208.png
bottom of page