Familiar, a startup in Y Combinator's S26 batch, says its AI dubbing keeps an actor's voice, lip movements and scene audio intact across 30 output languages. Its API documentation describes fewer.
The company's October 30, 2026 announcement calls the underlying technology a Multimodal Localization Decision Model, or MLDM. That is company positioning, not demonstrated superiority. The reporting behind this article turned up no outside benchmark or customer data against which to test it.
What the product does
According to the announcement, Familiar localizes video without stripping out the performance or the sound design. The original performer's voice, lip movements and surrounding scene audio are meant to survive the translation.
The company's homepage and studio-platform guide advertise 30 output languages and 99 source languages. The guide was published October 30, 2026 and updated November 1.
The language count doesn't add up
Familiar's API introduction and language documentation describe 24 languages, any-to-any. Two first-party figures, then, and no explanation for the gap in the materials reviewed. Buyers should not assume every advertised language is reachable through the API.
The number has moved before. Founderland reported on August 28, 2026 that the product was announced August 16 and supported 25 languages at that point. The original August announcement could not be located, so that date rests on Founderland alone.
Live dubbing is not available yet
August coverage and YC promotion described real-time translation. Familiar's guide, updated November 1, lists live dubbing as "in the works."
Real-time dubbing should not be treated as generally available, and the company did not confirm otherwise in the materials reviewed.
A small team, with an unsettled headcount
Sources disagree on size. The YC-derived directory startups.fru.dev reports eight people. Some Familiar article footers say seven; another says eight. The company's own team page names seven operating-team members and two advisers.
A crowded field
Familiar's pitch, preserving the performance rather than replacing it, overlaps with several established players.
ElevenLabs launched Dubbing v2 on May 28, 2026 (updated October 5), advertising original-performance conditioning and more than 90 languages. Its documentation, retrieved November 8, describes editing in the user interface, though transcript editing and regeneration through the API is limited to enterprise customers. The same documentation says v2 audio supports at most stereo rather than preserved 5.1 surround, and that live dubbing is unavailable. That complicates any blanket claim, including Familiar's framing, that competitors offer no editing or control.
HeyGen, in a guide updated November 3, documents voice-preserving translation, lip-sync, glossary controls and a Precision mode meant for cuts, side profiles and hands near faces. It also acknowledges that sound effects are removed and on-screen text is not translated in that workflow. Scene audio is where Familiar is positioning itself against that approach.
YouTube reported on May 15, 2026 that it offered auto-dubbing in 27 languages, with Expressive Speech in eight. Those counts are nearly six months old. On September 23, it announced a live auto-dubbing pilot planned for early 2027, and said more than 40% of live watch time comes from outside creators' home countries.
Flawless said its TrueSync tool won an HPA innovation award on February 20, 2026, and it emphasizes retaining original performance, emotion and timing. Lightcast announced AI subtitles and dubbing in its Media Cloud covering more than 35 languages, though the page retrieved carried no publication date.
Not to be confused with
Two similarly named companies turned up in the research. Familiar Labs, formerly MetaVoice Labs, is a San Francisco conversational-AI company led by Siddharth Sharma and Vatsal Aggarwal; BMW i Ventures discussed it in August 2026. Crunchbase also describes a familiarlabs.ai business offering persistent AI collaborators, shared workspaces and training. Neither is the dubbing company covered here.
What's missing
The reporting found no funding amount, publicly posted pricing, or named customers for Familiar, and no comment from its founders on the launch.
What to watch
Two things will show how much of the pitch holds up. One is whether the 30-language claim and the 24-language API documentation converge. The other is live dubbing, which Familiar lists as in the works while YouTube targets a pilot in early 2027.
