Turn your existing slide decks into presenter videos — in bulk

A single MP4 per deck. Each slide becomes a clip with the slide on one side and a digital presenter narrating it on the other, synced to the final audio. Start and end timestamps are recorded per page so viewers can jump to a specific slide.

Yes, on the self-hosted deployment. The whole pipeline — slide rendering, text-to-speech, lip-sync and final encoding — runs in a Docker stack on a machine you control. No deck, face clip or voice sample is transmitted to us. The managed cloud option, by contrast, stores assets on our infrastructure.

One Linux host with an NVIDIA GPU of at least 8 GB VRAM, plus Docker with the NVIDIA container toolkit. Text-to-speech runs on CPU; only lip-sync needs the GPU. A single GPU handles decks sequentially, and additional GPUs can be added to raise throughput.

Feature lists in this category overlap heavily, and the cloud platforms have larger model teams. The durable differences are deployment and unit economics: those products are cloud-only and bill per minute of output, while DocPresenter can run entirely on your own hardware where the marginal cost of another video is electricity. That matters most when you are converting hundreds of existing decks.

Yes. You upload reference audio, confirm in writing that you own the voice or hold consent to use it, and the voice is created through a provider that reviews it before it can be used. On self-hosted deployments you can also point narration at a speech model you host yourself. Cloning is deliberately not open to anonymous self-serve use, because verifying consent at that scale is an operational problem we do not yet solve well.

Registering a presenter requires an explicit declaration that you hold rights to that likeness. Terms prohibit uploading the likeness or voice of any person without documented consent, and specifically prohibit public figures, political figures and minors. Output videos are watermarked and carry AI-generation metadata. There is a reporting channel, and accounts that violate the policy are terminated.

Yes, and we recommend it. A draft narration is generated per slide from the extracted slide text, then presented for review. You can rewrite any page and regenerate the audio for that page alone, so a wording change does not force a full re-render.

Not yet. We are running a private beta with a small number of schools and corporate L&D teams while we measure render quality and cost per video. Leave your email and we will reach out when pilot slots open.