Skip to content

Slide Quest

Slide Quest is a live “side quest” for presenters: upload your deck, then summon fresh slides mid-talk with a quick voice description.

Code Demo Cerebral Valley

Core trick

Streams live speech through Deepgram transcription and extracts slide ideas via OpenAI Realtime to generate Gemini slides mid-talk.

Tags

voiceproductivity

Description

Slide Quest is a live “side quest” for presenters: upload your deck, then summon fresh slides mid-talk with a quick voice description. Gemini 3 handles the multimodal magic while Nano Banana Pro beams each new detour in crisp 2K, so every audience question becomes an instant visual story.
jakekinchen/SlideQuest★ 01 forksTypeScriptCreated Dec 6, 2025 · last push Dec 31, 2025

README excerpt

# Slidequest

Slidequest is a Next.js app that turns live speech, uploaded decks, and audience questions into AI-generated slides in real time.

Live demo: https://slidequest.vercel.app

It uses:
- Deepgram for live transcription
- Gemini for image and text generation
- OpenAI Realtime for idea extraction

The presenter gets a control surface with channels for exploratory ideas, audience questions, and uploaded slides, plus a separate audience view with QR-code joining and live Q&A.

## Requirements

- Node.js 20+ (22+ recommended)
- `npm` or `pnpm`
- API keys for:
  - `GOOGLE_API_KEY` (Gemini)
  - `DEEPGRAM_API_KEY` (Deepgram Nova)
  - `OPENAI_API_KEY` (OpenAI Realtime)

Create a `.env.local` file (not committed) based on these keys:

```bash
GOOGLE_API_KEY=your-google-gemini-api-key
DEEPGRAM_API_KEY=your-deepgram-api-key
OPENAI_API_KEY=your-openai-api-key
```

Office file uploads (`.ppt/.pptx/.key`) and server-side conversion are only enabled in local development (`NODE_ENV=development`) and require LibreOffice plus `poppler-utils` or ImageMagick installed on your machine.

## Local Development

Install dependencies:

```bash
npm install
```

Run the dev server:

```bash
npm run dev
```

Visit `http://localhost:3000` and:
- Start a presenter session from the main page
- Open the audience view (QR code or URL)
- Speak into the mic and watch slides appear in the exploratory channel
- Upload a PDF or images (and, in dev, PowerPoint/Keynote if conversion is available)
- Let aud

Team