Local-first video intelligence

TikTok Auto Project

A creator-owned workflow that analyzes long videos locally, measures candidate passages objectively, prepares short-form videos, and publishes only with explicit consent through TikTok's official APIs.

What the service does

Analyze locally

A long video is analyzed once on the creator's machine. Candidate windows are measured without generating hundreds of temporary clips.

Prepare videos

The local dashboard downloads approved sources and prepares vertical short-form clips for creator review.

Publish with consent

The creator connects their own account and explicitly starts exports through TikTok's official Content Posting API.

Objective candidate analysis

The first analysis engine produces raw, explainable measurements only. It does not ask an AI for a “viral score,” invent weighting, or send video content to a paid AI API.

silence_ratio

Share of the candidate duration classified as silence.

longest_silence_seconds

Longest continuous silent interval inside the candidate.

speech_density

Share of the candidate covered by locally detected speech.

words_per_minute

Transcribed words divided by actual active speech time.

hesitation_ratio

Configured hesitation expressions divided by analyzed words.

startup_latency_seconds

Time from the candidate start to its first detected speech.

LONG VIDEO → ONE GLOBAL ANALYSIS → CACHED TIMELINES → VIRTUAL WINDOWS → RAW METRICS

TikTok API usage

User control and safety