Does it fact-check the claims it finds in a video?

No. It labels claims, opinions, predictions, memories and questions but explicitly does not check whether assertions are true. Always verify consequential statements elsewhere.

Will YouTube link import work from a hosted deployment?

Not reliably. Fetching avoids the official Data API and often fails from cloud hosts. A successful deploy does not imply YouTube fetching works, so keep pasted SRT or VTT as fallback.

Can I try transcripts without configuring an API key?

Yes for browsing, pasting, viewing and exporting. Heatmaps, chapters and semantic search need a Gateway or TypeSafe key, and analysis starts automatically once a key is configured.

What data leaves my machine and who pays on a public link?

Pasted text goes to your app server and analyzed chunks go to the selected AI provider. There are no accounts or history, hourly caps are best-effort, and visitors can spend a deployed key.

How do I run it locally?

It needs Node.js 22 or newer. Clone the repo, run npm ci, copy .env.example to .env.local, and run npm run dev, then open http://localhost:3000. Without a key the sample or pasted transcript still opens and can be exported, but the heatmap, sections, and semantic search need a Jev provider.

How do I configure the analysis provider?

Two options: set JEV_PROVIDER=gateway with AI_GATEWAY_API_KEY, or JEV_PROVIDER=typesafe with TYPESAFE_API_KEY, leaving the unused provider's key empty. The app calls typesafe-ai/jev through AI SDK's experimental_evaluate, which is not a chat model call, so a random model name may not work. Restart the dev server after changing the variables.

How are timestamps handled for plain text without them?

Plain text with no timestamps gets times estimated at roughly three words per second and labels them as estimated in the interface. Any SRT or VTT exported from that text is also an estimate, and sentence times within a chunk are approximate; real video sync needs source timestamps, and the captions themselves can be wrong.

CommunityJev Tools & CLIs

Transcript Lens

Turkish Next.js app that classifies transcript passages with Jev instead of rewriting them, finding chapters, heatmap moments, claim types and topic matches while showing original wording.

See highlights

Open github.com

Stars: 18
Language: TypeScript
License: MIT
Published by: Community
Author: sensahin
Last verified: Oct 2026

sensahin/transcript-lens

User repository on GitHub

View on GitHub

YouTube transkriptlerini anlamına göre keşfedin. Türkçe arayüz, Jev analizi, altyazı dışa aktarma ve Vercel kurulum rehberi.

Forks: 2
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

YouTube transkriptini yalnızca okumak yerine anlamına göre keşfedin. Jev, metni yeniden yazmadan parçaları sınıflandırır; bölümleri, önemli pasajları ve aradığınız konunun geçtiği yerleri bulmaya yardımcı olur. Türkçe arayüzlü, Next.js tabanlı deneysel bir uygulamadır. Kendi bilgisayarınızda çalıştırabilir veya kendi Vercel hesabınıza dağıtabilirsiniz. YouTube bağlantısından altyazı almaya çalışır; SRT, VTT ve zaman damgalı metin yapıştırmayı destekler.

Quoted from the project's README. Read the full README.

Tagscommunity jevbest catalog-import

Works forJev for Classification

Used inJev for Software & DevTools

Languages: TypeScript

Highlights

Fetches captions or accepts pasted SRT, VTT and timestamped text with multiple export formats.
Derives chapter bounds and titles plus heatmaps without rewriting source passages.
Classifies passage types and retrieves topic matches as original quotes, not summaries.
Runs locally or on your own Vercel fork with Gateway or direct TypeSafe configuration.

Watch out

Captions come from unofficial player endpoints that can break or block cloud fetches, timing on plain text is estimated, Turkish headings can be weak, and public deploys can spend your key. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev-poly-crypto-demo

Research dashboard for Polymarket BTC five-minute Up/Down rounds that feeds market state to Jev class scores and paper-trades the edge.