Method 1: Extract first, translate with a free tool
If you only need this once or twice, the free path works fine. Pull the transcript first, then run it through a translator of your choice.
Steps:
- Paste the YouTube video URL into SubExtract's video captions tool
- Click Extract and copy the plain-text output
- Open one of the free translators below and paste the text in
- Pick the target language and copy the result
Which free translator to use:
- Google Translate — handles 130+ languages, accepts long text via the document upload (or paste in chunks). Best for speed and breadth of language support. Quality is good for major language pairs, weaker for low-resource languages.
- DeepL — generally the most natural-sounding output for European languages (German, French, Spanish, Italian, Dutch, Polish). Free tier caps at 1,500 characters per request, so longer transcripts need chunking. Worth it if accuracy matters.
- ChatGPT (or Claude) — context-aware translation. You can prompt it to "translate this transcript to Spanish, keep the conversational tone, preserve speaker turns" — and it will. Best when the video has slang, technical jargon, or tone you want preserved. Free tiers exist for both.
The free path's main downside: you're juggling two tools, and any timestamps or SRT formatting in the source get mangled when you paste through a translator.
Method 2: Translate during extraction (SubExtract Pro)
SubExtract has built-in translation on the Pro tier. Paste the URL, pick the target language, click Extract — the transcript comes back already translated, with timestamps and SRT formatting preserved.
Steps:
- Paste the YouTube video URL into SubExtract's video captions tool
- Toggle Translate and pick your target language
- Click Extract
- Copy the translated text or download as SRT
What you get over the free path:
- One step instead of two (no tab-hopping, no copy-paste)
- Works on long videos in a single request — no chunking around free-tier limits
- SRT timestamps stay intact, so you can drop the file straight into a video editor in the new language
- Consistent terminology across the whole transcript (free tools often re-translate the same phrase differently across chunks)
This is a built-in feature, not a workaround — it's the reason translation is part of the tool's Pro tier.
Comparison: which path fits which job
| Use case | Best path | |---|---| | One-off, casual reading of a single short video | Free — Google Translate or DeepL | | You want tone and jargon preserved (interviews, technical talks) | Free — ChatGPT or Claude with a custom prompt | | Translating subtitles for a video edit (need clean SRT) | Pro — built-in translation preserves timestamps | | Long video (1+ hour) | Pro — no chunking, single request | | You translate transcripts weekly | Pro — saves the copy-paste-juggle every time | | Multiple target languages from the same source | Pro — one extraction per language, no source re-translation | | Low-resource language (Tagalog, Swahili, Welsh) | Free — ChatGPT/Claude tend to handle these better than dedicated MT |
For a single video you watched and want to read in your language, the free path is honestly fine. For anyone doing this as part of a workflow — content repurposing, language learning at volume, subtitle production — built-in translation pays for itself in saved time.
Frequently asked questions
Which languages are supported? Google Translate covers 130+ languages and is the broadest free option. DeepL covers ~30, mostly European, and is the most natural for those. ChatGPT and Claude can handle most languages but quality varies — major languages are excellent, niche ones are hit-or-miss. SubExtract's built-in translation supports the same major languages as Google Translate's API.
Which method gives the most accurate translation? For European language pairs (English ↔ German, French, Spanish, Italian, Dutch), DeepL is usually the most natural-sounding. For everything else, an LLM (ChatGPT, Claude) often produces the best output because it understands context — sarcasm, idioms, technical terms. Built-in translation in SubExtract is comparable to Google Translate in quality but wins on workflow speed.
Are there length limits? Yes, on the free path. DeepL's free tier caps at 1,500 characters per request — long videos need chunking. Google Translate accepts longer pastes but truncates very long text. ChatGPT handles 50k+ words in a single message but you may hit context limits on the free tier. Pro built-in translation runs server-side with no chunking — works on multi-hour videos in one request.
Does translation preserve formatting (timestamps, SRT)? Free path: usually no. Pasting an SRT file into Google Translate or DeepL tends to break the timestamp lines or shuffle them. ChatGPT can preserve formatting if you prompt it to, but it's fragile on long inputs. Built-in translation preserves SRT structure cleanly because it translates the text spans, not the timestamp wrappers.
Can I translate without signing up? For the free path, yes — Google Translate and DeepL work without accounts. SubExtract's basic extraction has a free tier, but built-in translation is on the Pro tier and requires an account. If you only translate occasionally, the free path is genuinely fine.
What if the original transcript has errors? Auto-generated YouTube captions are roughly 95% accurate for clear English speech, lower for accents or technical jargon. Errors in the source carry through any translation — neither free tools nor Pro translation can fix what the original got wrong. For high-stakes uses, proofread the source transcript first, then translate.