How much does SlideTalk cost?

Three steps. 1) Drag a .pptx or .pdf into SlideTalk and say in one sentence how you want it presented. 2) Pick the target platform, narration mode and voice; AI streams the narration, and you can fine-tune it page by page in the studio. 3) Render in one click, and the Publish Center produces the cover, title, description and tags. No camera, no editing skills required.

Windows is not supported. Mac, iPhone and iPad are the same universal app, all three on the App Store (China mainland, Hong Kong and Taiwan storefronts), requiring macOS 15.1+ / iOS 17+ / iPadOS 17+ respectively — every device on the same Apple ID shares one credit wallet. A HarmonyOS version is still in planning and has not been released. On Mac it integrates deeply with drag-and-drop import, Spotlight search, the ⌘K command palette and native AVFoundation rendering.

It supports .pptx, .ppsx, .pdf and PNG / JPG / HEIC / WEBP images, dragged in from Finder, WeChat, Mail, AirDrop — anywhere. PDFs are rendered page by page into high-resolution images, which gives the most faithful layout. PPTX layout is approximated in place for text, images, shapes and tables, but gradients, charts, SmartArt and native PowerPoint animations aren't supported yet. Legacy .ppt / .pps and Keynote's .key can't be parsed directly — re-save them as .pptx or .pdf in PowerPoint or Keynote first.

Download is free, and importing, editing, voice preview, rendering and export (480p–4K, any length) never cost anything. Only AI generation is metered in credits: narration 1 credit/page, voiceover 1 credit/10 characters, AI covers per image — a 10-page video is about 352 credits (roughly ¥3.5). You get 600 free credits on sign-up (about 1.7 videos) and top up as needed: ¥6 / ¥28 / ¥68 / ¥198 packs, with the first purchase doubling your credits. There are no capability walls — no resolution, length, cover or title limits; the only gate is a brand watermark on exports, permanently removed by any first purchase. Credits never expire, and refunds are reversed automatically per Apple's rules.

About 352 credits — roughly ¥3.5 — covering AI narration and voiceover; the 600 welcome credits cover about 1.7 videos. Every AI run shows its projected cost and freezes it first; when your balance runs low generation pauses and the top-up sheet appears, then resumes from the checkpoint with no double charging.

No. The 2026 rewrite routes every AI call through SlideTalk's own gateway (dual-channel: Volcengine pay-as-you-go primary, Agent Plan fallback), so the client neither needs nor allows API keys. For narration, covers and vision understanding, only the necessary content of the relevant pages is forwarded through the gateway; the gateway stores no user content.

No sign-up, no login. The app uses Apple's AppTransaction to create an anonymous identity bound to a credit wallet (shared across every device on the same Apple ID), and 600 welcome credits arrive on first launch. The developer only ever sees a random account identifier — not who you are.

Four narration modes: solo narration, two-person interview, teacher-student Q&A and news anchor. Two-person modes automatically assign two different voices so the video feels like a real conversation. Voices come from Volcengine Seed-TTS 2.0 — 42 Chinese and 16 English across Lecturer/Education, Business/Broadcast, Warm/Lifestyle and Specialty/Trending — with 0.5x–2.0x speed control. All of them are free to use and previewing a voice costs no credits.

Yes, and both are automatic. Subtitles are split from the narration into CJK-safe short lines and burned into the frame, with a .srt file exported alongside. There are 6 scene animations — highlight, zoom, spotlight, arrow, underline and bracket — and AI reads your slide to pick the region worth emphasizing, then uses on-device OCR to snap it onto the real text lines; you can also drag things around in the animation editor. On top of that there are Ken Burns camera moves and cross-fades between pages, and long videos can be split into segments by page and exported separately.

The copy of your deck, slide images, narration, audio, video and covers all live inside the app's sandbox container (project files plus a SwiftData database), and rendering is done locally by AVFoundation — uninstalling the app deletes them all. Only the AI steps go online: during narration the text of the relevant pages (and the slide image if vision understanding is enabled) is forwarded through our gateway to the model provider, and voiceover sends the text to be read; the gateway stores no user content. Repeated lines play back instantly thanks to a local audio cache.

AI generation (narration, voiceover) needs the internet because it runs through our gateway. Scripts and assets you've already generated can be edited offline, and video rendering is entirely local — you can export without a connection.

The current version exports a complete release bundle — video, subtitles, cover, title candidates, description, tags and chapter timestamps — which you paste into the platform's back end to publish. Direct upload to B站 / YouTube is on the roadmap.

Creators who need to publish consistently but don't want to be on camera or can't edit: B站 knowledge creators, YouTubers, finance and science bloggers, course instructors and online teachers, corporate trainers and product managers. If your content can live on slides, it can become a finished video.