Hugo He
Work / SpeechLab

AI Dubbing Made Easy

Redesigning SpeechLab’s dashboard and editor for a clearer, faster dubbing workflow.

Role
Product Designer
Team
Design Manager, PM, Engineer, Founder, QA
In short

Dubbing meant too many steps, and many users left before finishing. I redesigned the dashboard and editor so: uploading starts transcription automatically the next step is clear at each stage teams can dub in batches Dubbing completion rose 13%. The full flow took 30 seconds less on average. Individual users needed 2 fewer clicks per task. Teams start batch dubbing with one setup.

Who it's for

Creators need a clear path.
Enterprise teams need efficiency at scale.

Creators dubbing their content

YouTubers, podcasters, TikTokers, and filmmakers need a clear path to a finished dub.

Enterprise teams handling hundreds of files

Enterprise teams, including Pearson, DeepLearning.AI, and iHeartMedia, need to handle large volumes of content efficiently while keeping control over transcript review.

The problem

Too many steps, too little guidance

Users had to start each stage and find the next step themselves.

Users dropped off throughout the dubbing flow

Hotjar funnel from registration to dubbing, before the redesign.

The redesign

A more intuitive dubbing workflow for creators and enterprise teams

Automatic transcription, guidance as work progresses, and batch actions for teams.

Upload, and transcription starts automatically

Choose the source language once and confirm the upload to start transcription. When it finishes, the project row expands to show the next step.

The next step comes into view

Start a translation and its column comes into view. When it finishes, the editor brings you to the action for creating a dub.

One setup for the whole batch

Select projects in the dashboard and create dubs in batches. Track each result as it finishes.

Research

What was behind the low completion rate?

The funnel showed where users stopped. Research pointed to two recurring problems: redundant actions and unclear next steps.

A still from a recorded usability session: the old dashboard in a browser, mid-upload, with the participant’s camera tile blurred at the top right.
Usability testingGave creators three tasks and noted where they got stuck or hesitated.
A Hotjar mouse-movement recording wandering across the dashboard.
Behavioral analyticsReviewed Hotjar funnels and session recordings to see where users dropped off.
A blurred Slack channel shared with an enterprise client.
Learning from enterprise clientsDiscussed enterprise workflows in calls and shared channels.
Design challenge

How might we help more users finish a dub with less effort?

Fix 1 · A shorter path

Start transcription on upload

It wasn't the first place that I went. I was checking if there was anything else I needed to click.

Creator, usability test
Before The separate Transcribe step was easy to miss
The old path: Dashboard, Upload, Click Transcribe, Open project, Editor.
After Transcription starts automatically on upload
The new path: Dashboard, Upload, then automatic transcription along a dashed line to Open project and Editor.
Fix 2 · A visible next step

Make progress and next steps clear

I don't know what to do next.

Creator, usability test
Before The next step was hard to find
The old editor, twice: a magnifier on the small Add a translation link in the column header, and the same screen after clicking it, unchanged.
After The next action comes into view as each stage finishes
Three editor states in a row: the filled Add new translation button, the translation column processing, then the dub column with Create your first dub.
Fix 3 · Make it scale

Dub in batches from the dashboard

I just want to process them in one go.

Enterprise client
Before Open each project in the editor to dub it
Diagram: four projects in the dashboard. Each one is opened in the editor for its own setup and produces one dub.
After Set up once for the whole batch
Diagram: four projects selected in the dashboard share one setup, and each gets its own dub.
Impact

After launch, more users finished with less effort

+13%

Final dubbing completion, one month after launch.

0.5 min

Average time saved across the full dubbing flow.

2 clicks

Saved per task for individual users.

1 setup

Starts translation and dubbing for selected files.

Learnings

What I took from the project

Work with constraints

I weighed user value against build effort, then planned changes the team could deliver in stages.

From system to detail

The full journey showed me where to focus. I refined buttons, states, and transitions to guide users forward.

Lower the language barrier

I want to use AI to help people do things that used to be out of reach.

Hugo HeNew York