Slide Talk Video Light

Turn a finished PDF slide deck into a synchronized, narrated MP4 video — automatically, with no manual video editing. Bring your own per-slide speaker script, or let the skill draft one from your slides for you to approve.

The text-to-speech needs no API keys and adds no extra cost: everything is included in your Capafy price.

What it does

You provide:

1. Slides — a finished PDF, up to 100 slides per run.
2. Narration — a per-slide speaker script, as structured YAML or plain text. Or nothing at all: if you only have slides, the skill drafts the script for you and asks you to confirm or correct it before rendering.

The skill then renders each slide to a high-resolution image, generates speech for each slide with built-in text-to-speech, assembles image + audio into one synchronized segment per slide, and concatenates everything into a single validated MP4.

Highlights

• No editing, no keys, no extra cost — no video editor, no paid voice API, no separate subscription.
• Reliable output — 1080p H.264 / AAC MP4, decode-validated before delivery. A file that merely exists is not enough; the final video is actually checked for a valid video stream, audio stream, and plausible duration.
• You stay in control of the script — narration lives in a readable YAML or text file you can review, version, and correct. No script yet? The skill drafts one from your slides and renders only after you have approved it.
• Per-slide artifacts — one audio file and one video segment per slide make retries and fixes easy.
• Pronunciation control — map abbreviations and brand names (e.g. "API" to "A P I") so they are read out naturally.
• Language voices — English, German, French and Chinese (Mandarin) are tested end to end; you can also name any other Edge TTS voice explicitly (see Limits below). The agent replies in the language you write in.

Typical use cases

• Training and onboarding videos from existing decks
• E-learning and course modules
• Internal communication and process walkthroughs
• Product or sales presentations as shareable video
• Quick video prototypes from slides you already have

What you get back

• the final narrated MP4
• a per-slide timing report (duration per slide and per segment)
• a technical build summary

Limits and requirements

Read this before you buy — these are hard limits, not preferences.

• PDF only. PowerPoint files are not accepted here. Converting them needs LibreOffice, which is not available in the hosted environment. Export your deck to PDF first — that also gives the most reliable layout.
• Up to 100 slides per run. The check happens before any rendering starts, so an oversized deck fails immediately instead of part-way through. Larger decks work well in parts: split them and build each part separately.
• A speaker script is needed — yours, or one drafted for you. If you bring a script, the skill narrates exactly what you wrote. If you only have the deck, it drafts a per-slide script from your slides and waits for your approval before anything is rendered. Slide bullet text makes poor narration, so scripts are written as spoken language.
• Narration and slides must match one-to-one. The counts are checked and the run stops if they differ, so a shifted or missing block is caught before rendering.
• Slides are static. PowerPoint animations, transitions, and embedded video are not carried over.
• An internet connection is required for the text-to-speech step.
• Verified languages: English, German, French and Chinese (Mandarin). All four are tested end to end, both locally and on Capafy itself. The underlying voice service offers many more and you can name a voice explicitly, but for anything beyond these four, run a short test deck first before committing a large one.

Data handling

The per-slide narration text you provide is sent to Microsoft's speech service to synthesize the voice track. No account or API key is involved, and nothing else leaves the environment — not your slide images, not the finished video. Please do not upload personal or confidential content.

Good to know

• The built-in voices produce clean, clearly narrated videos. They are not studio-grade or fully human-like voice cloning. For premium voiceover, use a full-scale presentation-video workflow instead.
• A typical run takes a few minutes. Time scales with the number of slides, so a 100-slide deck takes considerably longer than a 10-slide one.