Speaker Selection

Speaker Selection

Help users quickly find the perfect voice for their use cases, using data, among the vast array of AI voices.

Help users quickly find the perfect voice for their use cases, using data, among the vast array of AI voices.

Help users quickly find the perfect voice for their use cases, using data, among the vast array of AI voices.

My Role:

My Role:

My Role:

Product Designer

Product Designer

Product Designer

Timeline:

Timeline:

Timeline:

2024 Jun - Aug

2024 Jun - Aug

2024 Jun - Aug

Tools Used:

Tools Used:

Tools Used:

Figma, Stable Diffusion

Figma, Stable Diffusion

Figma, Stable Diffusion

01

Background

01

Background

01

Background

As AI voices continued to be updated on the voiceover tool Genny, I was in charge of creating and managing speaker avatar images, names, sample audios, and various metadata tags. During this process, I began questioning whether the current Speaker Selection modal was truly offering a good user experience. Since voice selection plays a key role in an AI voice tool, I successfully proposed and led a renewal of the feature to improve the overall UX.

As-Is

Lack of structure in the side navigation bar mixes technologies and features, and an incomplete filtering system fails to effectively sort speakers, resulting in a poor user experience.

To-Be

Design a speaker recommendation system based on collected user data and implement an advanced, accurate filtering and sorting system, enabling users to quickly find the speakers they need.

02

Key Values

02

Key Values

02

Key Values

Personalized Search Based on Use Cases

Using the data from the Onboarding Survey users complete when signing in to Genny, I matched a relevant Use Case from the 6 categories to each user.

Based on this, I designed a personalized search logic that recommends a speaker that fits the user’s specific use case.

Provide Data for Accurate Search

Provide Data for Accurate Search

Provide Data for Accurate Search

Update outdated speaker data, including age, style, and use case, by breaking them down into more detailed categories.

Collect and process data to provide users with more accurate information, enabling more effective searches.

Update outdated speaker data, including age, style, and use case, by breaking them down into more detailed categories.

Collect and process data to provide users with more accurate information, enabling more effective searches.

03

Features

03

Features

03

Features

Feature #1

Voice Library

Separated Voice Library and My Voice to help users stay focused on their specific goals

Feature #2

Trending & Badges

Built to help users stay updated with trends and reliable information,
offering a UX that makes it easy to choose the right speaker for their needs.

Feature #3

Alternative Suggestions & Coach Mark

Since this was a major renewal, the focus was on designing

a friendly and intuitive experience that guides users smoothly and keeps them engaged

Feature #4

Mobile Compatibility

Mobile-optimized UX ensuring the same level of convenience and functionality as on desktop,

with efficient navigation and interaction design tailored for smaller screens.

04

Retrospective

04

Retrospective

04

Retrospective

It was hard to convince the team to redesign the Speaker Selection UX. While everyone agreed on the need for the renewal, considering the company's current situation and resources, there was not enough capacity to turn the rapidly evolving AI features into an AI Video Editor for Genny. However, from the moment I joined the team, and as I saw the growing number of voices and received user feedback, it became clear that the renewal was inevitable. With the continuous customer service support and the addition of features like Voice Cloning and Collectible Voice, the Voice Selection became more complex, and it seemed like we could no longer postpone. So, I first conceptualized the screen and prepared the renewal plan I had in mind, then, together with the PM, we discussed possible directions considering the available resources and time.


Although the UI/UX adjustments were relatively straightforward, restructuring the voice data and modifying the data architecture demanded considerable time and effort from the engineering and UX teams. Gathering scattered data in Mixpanel and updating outdated voice data was a lengthy process. This project made me realize that UX is not only about proposing a flow but also about carefully designing the process of determining which data to present to users.