Three paths from audio to captions. Each one works. Each one costs you differently. In time, money, or quality. What you actually get depends on which you pick.
A caption track is a separate file attached to the video, not hardcoded over the image. The viewer toggles it with the CC button; YouTube's algorithm reads it as a discoverability signal. Each video can have multiple caption tracks, one per language.
Path 1. Manual typing in YouTube Studio
YouTube Studio → Subtitles → Add language → Type manually.
Opens a timeline editor. You listen to the audio, type what you hear, set the start and end time of each segment.
Time: 4 to 6 times the video's length per language. A 60-minute episode means 4 to 6 hours of work if you're doing it solo. Two languages (HR + EN) adds up to 8 to 12 hours.
Cost if you outsource: a professional captioner in Croatia charges roughly 8 to 15 €/min of input audio, depending on turnaround and content complexity. 60-min × 12 € ≈ 720 € per language. HR + EN therefore jumps to ~1440 €.
Quality: the highest possible. The human ear catches accent, irony, proper nouns.
When it's worth it: premium productions, branded content where you cannot afford mistakes, short but important videos (e.g. an official ad).
Path 2. YouTube auto-captions
Upload a video → YouTube automatically generates a caption track in the audio's language. You can enable auto-translate for a second language (English from Croatian, etc.).
Time: 0 hours of direct work. YouTube processes it within 30 minutes of upload.
Cost: 0 €.
Quality (impression, not WER measured):
- Croatian auto-captions handle clean narration with a single speaker decently. Proper nouns drop, code-switching breaks them, numbers are a lottery.
- English auto-translate from Croatian is grammatically passable, but names aren't translated consistently, idioms get lost, technical terminology often drifts off-topic.
When it's worth it: personal vlogs, experimental content, channels where viewers don't usually toggle captions on. Something is better than nothing. But not for an international audience.
Path 3. Automated pipeline (Titlomat)
You connect your channel via OAuth, and we generate Croatian and English caption tracks for every new upload and push them back to the video.
Time: 30 seconds to connect, 0 for every subsequent video. You can review before publishing if you want.
Cost: from 19 €/mo for 5 hours of video material. Pricing.
Quality: Phase-0 baseline across 3 channels yields a WER of 1.85% to 12.48% depending on content (see: Why Whisper makes mistakes on Croatian). A per-channel custom dictionary cuts errors by 3+ percentage points further.
When it's worth it: regular channels (≥2 episodes/mo), podcasts and interview formats, channels that want to grow beyond the Croatian-speaking audience.
Comparison table
| Path | Time per 60-min episode | Cost | WER | Scales |
|---|---|---|---|---|
| Manual (you) | 8 to 12h for HR + EN | 0 € | ~0% | No |
| Manual (outsourced) | 1 to 2 days lead time | ~1440 € (HR+EN) | ~0% | Expensively |
| YT auto-captions | 30 min wait | 0 € | variable | Yes |
| Titlomat | ~5 min processing | from 19 €/mo | 1.85% to 12.48% | Yes |
How the YouTube algorithm sees this
YouTube recommends captions as a discoverability signal. Videos with captions rank better in countries whose language is only covered by the caption track. Concretely: a Croatian podcast with an English caption track becomes a recommendation candidate for viewers in Germany, the UK, and the US.
Our internal methodology for measuring how many more viewers this actually brings: here.
What to pick
- One important video a year? Manual, outsourced. Premium quality justifies premium pricing.
- A channel that isn't primarily about growth? YT auto-captions. There's no better free option.
- A regular channel with international-audience ambitions? Automated pipeline.
Titlomat covers the third category. Connect a channel, the next episode automatically gets Croatian and English captions, WER and reach lift are visible in the dashboard. Try it free. The first week is on us.
