Your video. A transcript you can search.

Turn video and audio you own or are authorised to use into a transcript with timestamps. Find the moment you need and export original-language SRT or VTT.

  1. Your video or audio
  2. Transcript with timestamps
  3. Search
  4. SRT / VTT

See a real transcriptSee how VTI works

Personal testing · closed beta

VTI · 2026-10-29

A real recognition result, kept as it was

Saved output from a local Whisper test on a 6.7-second video. The input uses synthetic speech, not a customer recording. Opening this page starts no processing.

vti-olive-oil.mp4Saved recognition · English

local-whisper-tiny · 6.7 s · synthetic speech

  1. Add two tablespoons of olive oil to the hot pan.

  2. Olive oil helps the past and not stick together.

These are the whole-second start times shown in the test. Subtitle end times are not inferred here.

The recogniser wrote “past and” where the source said “pasta”. Review automatic transcripts before publishing. One short test does not establish accuracy for other files or long recordings.

Try the separate search and subtitle exercise

Explore this original, prepared text example. Search the three passages or download their subtitles. No recording is uploaded and no AI runs here. These downloads contain the exercise text, not the recognition result above.

VTI field notes · prepared textSample · English
Local text matching only. The product’s semantic search runs in the private workspace.
  1. Keep a transcript so you can return to the original words.

  2. Use timestamps to find the moment that matters.

  3. Export subtitles as SRT or VTT for your editing workflow.

3 passages · 12 seconds of illustrative timing

What you get

The closed upload workflow is built around these outputs:

When VTI fits your workflow

We complement YouTube, Google and Twitch rather than compete with them. When a platform's own tools already solve your need well, use them.

VTI is for authorised media you need to transcribe, search and export for editing or publishing. Translation is a separate, configuration-dependent capability.

We do not claim that any particular platform lacks a feature. What a platform supports depends on region, language, content type and date, and it changes.

How it works

Four steps, in one place:

Supported inputs

The upload route supports these formats; check the workspace for current limits:

Supported languages

These language codes are supported by the product contract. Availability and quality depend on the configured processing path. Interface languages are a separate setting; translation is not enabled by selecting one.

Pricing

Plans and quotas are read from the live configuration of this deployment:

Paid plans are not published yet. Early access is coming soon.

Rights and limitations

VTI works on media you own or are authorised to use. It does not bypass platform protections and does not extract media you are not entitled to.

Automatic transcription and translation can contain errors. Review the output before publishing. Quality varies with audio clarity, accent, overlapping speech and subject-specific terminology.

Questions

Is VTI a replacement for YouTube's own translation?

No. It complements it. Where a platform's own tools already cover your language pair and content type, use them — that is usually the simplest option. VTI is for media you upload yourself, subtitle exports for editing and publishing, and one pipeline over transcripts, translations and search with citations.

Which languages can VTI transcribe and translate?

The product contract includes English, Portuguese, Spanish, German, French and Russian. The active recognition and translation paths determine availability. This does not promise that translation is enabled in the closed test.

Can I use a video I do not own?

No. You must own the media or have permission to use it. VTI does not bypass platform protections or extract inaccessible media.

What can I export?

Original-language SRT or VTT from a completed transcript. Translated exports require a configured translation path; speaker labels are not guaranteed.