Loading Project Media
- Year
- 2026
- Type
- Product
- Status
- Archived
A low-latency translation product for cross-border commerce. It was conceived to turn natural spoken language into a target language with as little setup and interruption as possible.
Project visual exploration — concept artwork, not a live product screenshot.
(Overview)
A real-time translation workflow for cross-border ecommerce and livestream selling. Substantial workflow development led to validation with US TikTok sellers, where adoption mattered more than technical feasibility.
Origin
Started around June 2026 after Gemini translation capabilities suggested a practical livestream workflow.
Archive reason
Validation with US TikTok sellers showed limited willingness to integrate APIs or additional tools. Some preferred lower-cost human operators even with imperfect English. I chose to archive the direction because the workflow change was not compelling enough, rather than because translation was technically impossible.
(Context)
The opportunity was to help cross-border sellers communicate during a live session. The product question was whether sellers would change their existing setup to use translation APIs and tools.
(Question)
How can a cross-border seller speak naturally and be understood almost immediately?
(Approach)
The workflow removes studio-style complexity: choose the source and target languages, begin speaking and let the translation keep pace. The interface gives the original and translated lines stable positions so the user can follow what was said without losing the current moment.
(Interaction)
The workflow explored source and target language selection and translated speech during a selling session. The portfolio sample uses a scripted English–Chinese sequence; it does not capture a microphone or connect to a translation service.
(Decision)
Technical feasibility does not establish willingness to change a buyer’s workflow. Validate the integration burden with the people who would operate it.
(Demonstration)
Play, pause or select a line in a scripted English–Chinese sequence. This browser sample does not access a microphone, capture speech or connect to a translation model.
- 01
Cross-bordercommerce
- 02
Low-latencyspeechtranslation
- 03
Naturalvocaltone
- 04
Minimalsetup
Haveaprojectinmind?
hi@dustinstudios.com