Presenters, voices and avatar video
A presenter is a character with a name, a face and a voice, and all three stay with them across every video. Avatar videos run to about two minutes. Uploading a voice recording shapes how a voice sounds — it is not voice cloning.
This applies to the formats that put a person on camera: the avatar video and the influencer review. Other formats carry music and on-screen text instead of a speaker.
How long can an avatar video be?
About two minutes, roughly 400 spoken words. Short is still the default — these are scroll-stopping clips, so unless you ask for something longer you get the usual few-second hook and payoff. Asking is all it takes.
A script longer than a single take is filmed as several takes from the same presenter and joined automatically. Three things follow from that: the joins are visible as cuts, the way an ordinary social edit cuts; the voice holds across them; and a longer video costs roughly proportionally more, because each take is generated separately. A script past the ceiling is refused with the number to cut to rather than being quietly shortened.
Can I clone my own voice?
No. You can upload an audio file in the presenter's voice picker, and it shapes how the generated voice sounds — its pitch, register, accent and pace. The words in the recording are never spoken back, and the presenter will not sound like the person who recorded it.
The honest framing is "give them a voice like this one", not "use my voice". What it does deliver is consistency: that presenter sounds the same in every video from then on. Any audio file works, a phone memo included.
There is no standalone text-to-speech feature and no way to supply a finished voiceover track. Speech is produced by the video model as it renders the clip. Music is separate and can be swapped on a finished video without re-rendering it.
Can the presenter look like me?
Your face can, but they will not sound like you. Upload a picture of a person — in chat, or kept in Brand & style as a brand asset — and ask for the presenter to be built from it. The generated presenter follows the face, look and styling in the picture. It is a strong likeness rather than a pixel-exact copy.
Which languages can a presenter speak?
Speech is supported in English, Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean and Russian. Content can be written in many more languages than it can be spoken in.
Outside that set — Hebrew, Arabic and others — it is handled automatically rather than failing: an avatar video is routed to a model that speaks the language but is limited to shorter clips and landscape or portrait framing, and longer narrated formats are rendered without spoken audio, with the narration carried by on-screen text. A chosen voice cannot be carried on those routes.
Influencer-review videos are always spoken in English, whatever language the rest of the post is in.
Is there an avatar library to browse?
No. There is no avatar marketplace, no library of ready-made avatars, no 3D avatars, no cartoon builder and no face customisation sliders. The presenters your account has already filmed with are collected under Influencers in Brand & style, and can be reused, renamed or given a different voice from there.
Do showreels have subtitles?
Every showreel carries its narration on screen, in any language, spoken ones included — most feed viewers watch with the sound off. Position, size, font and colours are set by the Overlay Text control on the post card, and changing them re-burns the existing video for free.