a guided tour

VoiceVideo

3 minutes · every feature
press F11 for fullscreen
VoiceVideo
a guided tour

VoiceVideo

Let's take a look around.

3 minutes · every feature

00What you can do here

From any topic to a published video.

01
Write
Claude generates the document from your topic.
02
Verify
Five-layer AI audit scores every claim.
03
Voice
ElevenLabs narrates in your custom voice.
04
Render
Browser turns slides + audio into HD MP4.
05
Publish
Direct push to YouTube, Notion, or GitHub.
06
Observe
Built-in cost optimizer and usage dashboard.
Section00 / 11
Time00:06 / 03:00
01Step one: pick the content type

Six formats to start from.

Every format gives Claude a different structure. An essay flows in paragraphs. A report has executive summary and recommendations. A debate gives both sides plus a verdict.

Essay / article
Report / analysis
Lesson / tutorial
Voiceover script
Debate / argument
Story / creative
Section01 / 11
Time00:18 / 03:00
02Step two: tell Claude what to write

Type your topic.

A single sentence is enough. You can also paste a source URL and Claude will read that page first. Or toggle web grounding to pull live search results.

Topic or title
The rise of AI-generated video content in 2025
Additional context (optional)
Focus on the verification problem — what happens when models hallucinate.
Section02 / 11
Time00:32 / 03:00
03Step three: dial in the voice

Tone. Length. Audience.

Three quick selectors decide how Claude writes. Engaging or formal. Short or comprehensive. General audience or specialist.

Tone
Engaging
Length
Medium
Audience
General
Section03 / 11
Time00:45 / 03:00
04Step four: Claude writes

Document appears in seconds.

The rise of AI-generated video content in 2025

In the space of eighteen months, AI video has gone from awkward novelty to production-ready tool. The shift is most visible in three places: marketing teams replacing agency reels with prompt-driven shots, educators generating multilingual lessons in minutes, and indie creators producing weekly series that would have demanded a film crew.

But the speed has outrun the verification. Models routinely fabricate statistics, attribute quotes to the wrong person, and confidently describe products that do not exist. The first wave of tools focused on generation. The second wave is starting to focus on something more important: trust.

This piece looks at where AI video creation is today, why the verification gap matters more than the model improvements, and what...

Section04 / 11
Time00:58 / 03:00
05Step five: refine in chat

Not quite right? Tell Claude.

The enhance chat sits below your document. Ask for a punchier opening, a different angle, more statistics. Claude rewrites the document in place.

Make the opening line punchier. Add a statistic about hallucination rates.
Done. New opening: "Forty-three percent of AI-generated videos still contain at least one fabricated claim. The race to fix that is now bigger than the race to make the videos themselves." Restated the next paragraph to follow on.
Section05 / 11
Time01:09 / 03:00
06Step six: the verification panel

Five layers, scored.

Before you publish, Claude scores the document on accuracy, bias, completeness, hallucination, and quality. Each layer gets a donut. The overall Trust Score sits at the top.

79
Layer 01
Accuracy
Pass
62
Layer 02
Bias
Review
83
Layer 03
Completeness
Pass
87
Layer 04
Hallucination
Pass
81
Layer 05
Quality
Pass
Section06 / 11
Time01:22 / 03:00
07Step seven: cross-check with a second AI

Two providers verify.

Optionally turn on a second judge: GPT-4, Gemini, or Mistral. Each verification layer runs in parallel. If they agree, your trust score gets the green badge. If they disagree, you see the gap before you publish.

Layer 01 · Accuracy
Agreement
79
Claude
82
GPT-4
Layer 02 · Bias
Disputed
62
Claude
84
GPT-4
Section07 / 11
Time01:39 / 03:00
08Step eight: generate the voice

Your custom voice, in seconds.

ElevenLabs Multilingual v2 reads the document in the voice you've chosen. Chunked streaming means a five-minute script takes under a minute to generate.

Section08 / 11
Time01:53 / 03:00
09Step nine: render the video

Pick a quality. Click create.

The browser turns your slides plus voice into an MP4. No server upload. No watermark on the video itself. Branded intro and outro come built in.

Standard 720p
HD 1080p
Premium 1440p
Ultra 4K
Section09 / 11
Time02:04 / 03:00
10Step ten: publish

Four destinations, one click.

YouTube auto-upload via OAuth. Notion database via webhook. GitHub commit via API. Or just download the MP4 and post it yourself.

YouTube
auto-upload
Notion
database sync
GitHub
commit
Download
MP4
Section10 / 11
Time02:17 / 03:00
11Step eleven: keep an eye on cost

Built-in observability.

Every API call is logged. The cost optimizer flags expensive prompts. Export the metrics as JSON for Grafana dashboards.

$0.12
total cost
5,128
input tokens
6,847
output tokens
12
api calls
Section11 / 11
Time02:30 / 03:00
your turn

VoiceVideo

Three generations are free on us. After that, paste your own Claude and ElevenLabs keys. Your usage, your bill, your data.

voicevideo.io   →

no install · no sign up · runs in your browser