DigiSelf Launches World’s Fastest Real-Time AI Avatars, Bringing Conversations Off the Cloud

New benchmarks show DigiSelf outpacing leading AI avatar platforms in response speed, with lifelike avatars that can hold live conversations and run completely offline

TOKYO, Oct. 21, 2026 /PRNewswire/ — Tokyo-based video generation AI developer EmbodyMe, Inc. today announced the global expansion of DigiSelf (https://digiself.tech), the world’s fastest real-time AI avatar platform, according to competitive benchmarks conducted by the company. DigiSelf turns a single photo into a lifelike digital person that can appear in videos and virtual meetings, speak multiple languages, and hold natural, real-time conversations.  The technology can also run completely offline on a user’s own hardware, without sending user data to the cloud.

In competitive benchmarks conducted by EmbodyMe, DigiSelf recorded the fastest response times among the leading real-time AI avatar services tested. DigiSelf began speaking incoming audio in a median of just 0.33 seconds, compared with 0.62 seconds for the next-fastest competitor, making it approximately 47% faster. In live voice conversations, DigiSelf began answering in a median of 1.17 seconds after a user finished speaking, compared with 1.94 seconds for the next-fastest competitor, representing approximately 40% less waiting time. These results highlight DigiSelf’s ability to deliver faster, more natural interactions with less delay between responses.  DigiSelf’s offline SDK is even faster, beginning to speak incoming audio in just 0.005 seconds when running locally on a user’s own hardware.

“We’ve spent nearly a decade developing real-time AI technology, with our products reaching more than two million downloads around the world,” said Issay Yoshida, CEO of EmbodyMe. “DigiSelf takes that experience to the next level. Speed is what makes an AI avatar feel like a real conversation instead of a series of delayed responses. Our benchmarks show that DigiSelf is faster than the leading platforms we tested, and we’re making that technology available online, in virtual meetings, and completely offline.”

Users create an avatar from a single photo and about 10 to 30 seconds of their voice, or from a video of about 30 seconds to 2 minutes. DigiSelf creates a dedicated AI model for each avatar to produce a highly realistic likeness that reproduces the person’s facial expressions and mouth movements, as well as their voice, speaking style, and mannerisms. Users can create entirely new videos simply by providing new scripts, without having to record or film themselves again. This makes it possible to quickly create and update video content without reshoots, saving time and production costs.

DigiSelf is designed for both personal and business applications, including marketing, sales, training, customer service, reception, and digital signage. Avatars can also be used in virtual meetings on platforms including Zoom, Google Meet, and Microsoft Teams. DigiSelf supports multilingual video and audio, making it possible to create versions of the same content for audiences around the world. The technology can also animate anime characters, animals, paintings, and other images.

From the Cloud to Completely Offline

DigiSelf offers businesses two ways to integrate its AI avatar technology: an online API for cloud-based applications and a local SDK that runs entirely on their own hardware.  The SDK works completely offline, with no user data sent to the cloud while it is running. Offline deployments are available through a one-time license with no per-minute usage fees.

DigiSelf avatars can respond in about 0.3 seconds through the API and in 0.005 seconds with the offline SDK, enabling real-time, back-and-forth interactions.

Already in Use in Japan

DigiSelf first launched in Japan in March 2026 and has already been used by broadcasters, public figures, and government. Bunka Broadcasting, working with Dentsu, used an AI avatar of Atsushi Tamura, one of Japan’s best-known comedians, to read news on his radio show. An AI avatar of announcer Keitaro Sunayama also filled in while he covered the Olympics.

Shizuoka Prefecture used a DigiSelf avatar of Governor Yasutomo Suzuki to open TECH BEAT Shizuoka, a technology event the prefecture hosted. The office of the late Kazuko Hosoki, Japan’s most famous fortune teller, has also used the technology to create AI videos and paid readings in her voice.

DigiSelf builds on EmbodyMe’s work in real-time AI face animation dating back to 2018. Founded in Tokyo in 2016, the company has launched seven products spanning social VR, real-time photo animation, virtual meetings, AI conversations, avatars, and SDK technology.

EmbodyMe’s apps have generated more than 2 million downloads in more than 200 countries and regions, with approximately 95% coming from outside Japan. The company has received the NVIDIA Inception Award and Microsoft Innovation Award, participated in the Techstars Music accelerator in Los Angeles, and holds patents in the United States, Europe, China, and Japan.

Availability and Pricing

DigiSelf is available today at https://digiself.tech. A free plan lets anyone try it, and paid plans, which include real-time conversations with your avatar, start at $29 per month, or $24 per month billed annually. Business and enterprise plans are available for teams. Developers can build with the DigiSelf API, with real-time avatar streaming starting at $0.10 per minute, and the offline SDK is available with a one-time license starting at $9,900.

About EmbodyMe, Inc.

EmbodyMe, Inc. is a Tokyo-based company specializing in generative AI, enabling user-generated video and image manipulation content. With a team of experts in deep learning and computer graphics technologies, EmbodyMe provides innovative solutions to today’s most pressing audio-visual communication challenges. The company boasts support from renowned investors, including Techstars, IncubateFund, and DeepCore (a Softbank AI-focused fund). EmbodyMe has previously released successful products such as EmbodyMe, a VR app, xpression, xpression camera, xpression chat, xpression avatar, and DigiSelf.

View original content to download multimedia:https://www.prnewswire.com/news-releases/digiself-launches-worlds-fastest-real-time-ai-avatars-bringing-conversations-off-the-cloud-302914962.html

SOURCE EmbodyMe