Marketing ops · Production
ElevenLabs showcases multilingual AI voice technology with NVIDIA ACE at Computex
The problem
Keynote presentations traditionally required live translators or subtitles to support multiple languages, and NPC interactions in video games were brief and scripted, limiting immersion.
Workflow diagram · grounded in source
1
AI voice clone creation
ai_action
“He was able to create the AI voice on ElevenLabs in under an hour with just seven minutes of recorded audio”
2
Multilingual voice generation
ai_action
“enabling multi-lingual keynotes that preserve the speaker's voice traits using its proprietary AI speech and voice cloning technology”
3
Multilingual keynote delivery
output
“Jensen Huang narrated several chapters of his Computex keynote in both English and Mandarin using an AI voice”
4
Developer NPC microservices integration
integration
“Developers can integrate individual – or all – ACE NIM microservices directly into their products, tools, services or games and experiences”
5
Adaptive NPC voice interactions
ai_action
“The Covert Protocol tech demo, uses ElevenLabs voice AI to power those interactions”
Reported outcome
Jensen Huang narrated his Computex keynote in both English and Mandarin using an AI voice created in under an hour from just seven minutes of audio.
The partnership also enables human-like NPCs that react and adapt to gamers, praised by developers, journalists, and influencers for realism.
Reported metrics
time to create AI voiceunder an hour
Recorded audio required for voice cloneseven minutes
Reported stack
ElevenLabsNVIDIA ACEACE NIM microservices
Frequently asked questions
What did this team achieve with this AI workflow?
Jensen Huang narrated his Computex keynote in both English and Mandarin using an AI voice created in under an hour from just seven minutes of audio.
What tools did this team use?
ElevenLabs, NVIDIA ACE, ACE NIM microservices.
What results were reported?
time to create AI voice: under an hour; Recorded audio required for voice clone: seven minutes (source-reported, not independently verified).
How is this marketing ops AI workflow structured?
AI voice clone creation → Multilingual voice generation → Multilingual keynote delivery → Developer NPC microservices integration → Adaptive NPC voice interactions.