AI-powered document conversations

Your documents,
now speak back

Upload any PDF and have a real-time voice conversation with it. DocVoice embeds your content with AI and lets you — or your customers — simply ask questions out loud.

app.docvoice.ai/dashboard
Documents
API Keys
Embed
annual-report-2024.pdf
142 chunks · Uploaded today
Talk
DocVoice
Hello! Ask me anything about your annual report.
You
What was the Q3 revenue?
DocVoice
Q3 revenue reached $4.2M, up 18% year-over-year according to page 12.
50 MB
Max PDF size
< 1 s
First AI response
∞
Documents per account
Process

How it works

Three steps from upload to live voice conversation.

1
Upload your PDF
Drop any PDF document into your dashboard. DocVoice extracts, chunks, and embeds the text into a vector index — the whole pipeline takes under a minute.
2
Generate an API key
Create a key linked to your document. Optionally restrict it to specific domains. Copy the one-line embed snippet — it contains your key and server URL automatically.
3
Talk from anywhere
Paste the snippet into any website. A floating mic button appears. Visitors can speak or type, and DocVoice answers using only your document's content via Gemini Live.
Capabilities

Built for reliability

Everything you need to deploy a production-ready voice interface on your documents.

Voice
Real-time audio
Full-duplex streaming via Gemini Live. The AI listens, speaks back, and handles interruptions naturally.
Security
Origin-locked keys
Lock each API key to specific domains, or enable wildcard access for development. Revoke any key instantly.
Embed
One-line integration
A single self-contained script tag adds the floating widget to any site. No npm, no bundler required.
RAG
Pinecone vector search
Answers are grounded in your document's exact content via semantic retrieval — the AI will not hallucinate outside it.
Privacy
Per-account isolation
Every document and key is scoped to your user account. Other accounts cannot access your indexes.
Performance
Gemini 2.5 Flash
Native audio model with sub-second first-token latency. Streaming audio plays as it generates — no waiting.
Ready

Your first voice chat is minutes away

Free to start. Upload a document, generate a key, and talk to it. No credit card, no infrastructure.

DocVoice
1
2

By creating an account you agree to our terms of service.

2

Didn't receive it?