How to Get Bilingual YouTube Transcripts Side by Side

Comparing a video's original transcript against a translated version — for a language class, a localization review, or just understanding a foreign-language source — means lining up two transcripts side by side, not just reading each one separately. The plain exports (TXT, CSV, JSON, JSONL) don't carry timestamps you can line rows up by; for that you need either the API's per-segment data or SRT/VTT files, both covered below.

Two ways to get a second language

A video can have a second-language transcript two different ways:

A native track YouTube already has. Some videos ship with caption tracks in several languages, uploaded by the creator or auto-translated by YouTube itself. These are listed per video — no extra processing needed, and re-reading something you've already processed is free.
Auto-translation. If YouTube doesn't offer the language you want natively, Scriptube can translate to English with Google Translate (12 credits per 1,000 characters, Starter plan and up), or produce an auto-translated track in another language for 1 extra credit (also Starter+). If YouTube itself already provides the translation, it's free beyond the video's original 1-credit charge.

Check what's available first

Before requesting a translation, check which languages a video actually has natively — there's no point paying for a translation YouTube already provides for free:

curl https://scriptube.me/api/v1/videos/VIDEO_ID/languages \
  -H "X-API-Key: YOUR_API_KEY"

This returns the original language, any native tracks, and the available translation targets for that specific video.

Worked example

Say you have a video in Spanish (a placeholder scenario — swap in your real source) and want it next to an English version for a bilingual study sheet. First check /languages to see if YouTube already has an English track. If it does, request both with language=es and language=en — no translation charge either way. If it doesn't, set translate_to_english on your batch request and Google Translate fills the gap.

How to do it in Scriptube

Dashboard: submit the video, then use the language selector to pull the transcript in each language you need. The dashboard and its text/CSV export show plain text per language, not timestamps, so for a quick read-through just paste each language's text into its own column. For a timestamp-aligned comparison, use the API approach below, or export each language as SRT and line the two files up by timecode.

API: the cleanest way to get both languages in one response is exporting with language=all:

curl "https://scriptube.me/api/v1/transcripts/BATCH_ID/export?format=json&language=all" \
  -H "X-API-Key: YOUR_API_KEY"

To request English translation on the original submission:

curl -X POST https://scriptube.me/api/v1/transcripts \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "urls": ["https://www.youtube.com/watch?v=VIDEO_ID"],
    "translate_to_english": true
  }'

The export above (and the Google translation's translated_text) is plain text with no timing. If you want a timestamp-aligned comparison instead, call the transcripts endpoint with include_segments=true once per language — e.g. language=es and language=en — and align the two segment lists by each segment's start value: segment starting at the same time in language A sits next to its counterpart in language B. Note that only native and YouTube auto-translated tracks carry segments this way; the Google translate_to_english text has no segments to align.

Choosing a layout for the comparison

A two-column spreadsheet works well when both languages have roughly the same number of segments and similar pacing: pull each language's segments via the API and put one language in column A, the matching segment's text in column B, using each segment's start value as a shared key you can sort by. For a document-style comparison instead of a spreadsheet, export both languages as SRT and interleave them by timecode, or alternate paragraphs in a plain-text document, original line followed immediately by its translation.

Watch for pacing mismatches

Translated segments don't always map one-to-one with the original. A short phrase in one language can expand into a longer sentence in another, which shifts how segments line up over the course of a long video. If you're building something precise — a subtitling reference or a close textual comparison — check a handful of segments by hand against the original audio rather than assuming every row lines up perfectly by position alone.

A cost-planning example

Say you need 20 videos translated to English using Google Translate, and each transcript runs roughly 3,000 characters. At 12 credits per 1,000 characters, that's about 36 credits per video for the translation alone, on top of the 1 credit to process the video itself — call it 37 credits per video, or roughly 740 credits for the full set. Run the numbers for your actual videos with POST /api/v1/transcripts/estimate before committing, since transcript length varies a lot by video.

When native tracks are the better choice

If YouTube already has a native track in your target language, use it instead of auto-translation — it's free beyond the video's original 1-credit charge, and a native or creator-uploaded track is often more natural than a machine translation of the original language. Reserve auto-translation for the gaps native tracks don't cover.

Reviewing auto-dubbed releases

If you're reviewing a video that YouTube or a creator has auto-dubbed into another language, pair the dubbed track with the original using the is_original field to confirm which one is the source audio, then compare the dubbed transcript against the original side by side to check how closely the dub follows the source meaning.

A workflow for translators and reviewers

If a human translator or reviewer is checking machine translation quality, give them the side-by-side export rather than just the translated file alone — flagging a mistranslation is much faster when the original line sits right next to it than when a reviewer has to replay the video to check context.

Exporting for subtitle software

If the end goal is burning bilingual subtitles into a video rather than just reading the text, export both languages as SRT and import them as two separate subtitle tracks in whatever editing software you use — most subtitle and editing tools accept multiple SRT tracks on the same timeline without needing them pre-merged.

Where this is useful

Language learners use side-by-side transcripts for shadowing practice and vocabulary lists — see studying any video for language learning for the broader workflow. Localization and content teams use the same approach to spot-check translation quality against the source. For translation mechanics in more depth, see translating a transcript into any language.