Video Subtitling and Dubbing

Chances are good that you have a catalog of videos for product demos, webinars, training material and so on. This asset class poses its own set of challenges, and is often overlooked in the first pass at localization.

Most TMS platforms provide only rudimentary support for localizing video assets, for example, by translating SRT caption files that are paired with videos. Before AI captioning was available, video captions were typically authored by people, and then translations (subtitles) were created based on those captions by translators (which was very expensive, another reason companies would often put this off).

AI has become quite good at generating captions that require at most a modest effort to correct, and that caption can be fed into an AI+Human workflow.

There are several platforms that make this process largely automatic, and also integrate with popular video hosting platforms like Vimeo. One of my favorites is CaptionHub which provides an integrated solution that supports:

Automated or manual video ingestion and distribution from/to popular video hosts like Vimeo, Brightcove, etc. Automated distribution (but not ingestion) for YouTube.
Automated or AI+human translation for subtitles (translated captions) for over 100 languages
Automated or human voiceover (dubbing) based on subtitles
Automated delivery to popular video hosts
No code solution that is easy for non-technical employees to work with, and most often the people tasked with producing these assets are not coders.

The service is billed based on the number of minutes of video processed (video minutes times the number of languages targeted), which makes cost estimation straightforward. The integrations with video hosts also enable users to configure a “set it and forget it” workflow where videos are captioned, subtitled (or dubbed) by default.

A good way to think of CaptionHub is as a TMS (translation management system) for video content. It orchestrates the work of captioning and subtitling this content. Meanwhile, you can think of video hosts like Brightcove and Vimeo as content management systems for video. This is your central video repo. And lastly, you can think of services like Instagram, TikTok and YouTube as mechanisms for discovery and mass distribution.

Subtitling Versus Dubbing

From my perspective, whether to subtitle or dub (voice over) depends on the type of video content being translated. I personally prefer captions and subtitles because this solves for both language accessibility and hearing accessibility. It is also tricky to time voiceovers with the source language, whereas captions can be resized and can be displayed in a number of ways.

Inline Versus Out Of Band Subtitles

One decision you will also need to make with captions is whether to use inline or “burned in” captions that are embedded within the video, or to use out of band captions that are stored in separate text files (SRT is a popular file format). I generally recommend going with the latter because most video hosts support them, and some can also display them outside the main video window, which prevents captions from obstructing the video. This can also help with search visibility (one thing you can do is to link to the translation text in the landing page for the video).

For better SEO – it is generally good to include subtitle text in the page where the video lives. For example, if the video is embedded in a CMS page, display the subtitle text below. This will be crawled by search engines, and is also helpful to people are hearing impaired or just want to read the transcript.

Differentiating By Asset Type

Instructional Demos

Most companies have tons of instructional demos in their help center and related sites. My general advice with these is to use auto captioning and subtitling for long tail content, but for your “top ten” instructional videos, be sure to have human reviewers double check the auto generated captions for correctness. With tools like Caption Hub you can opt for AI only, AI + human or human only workflows. Subtitles are generally better than dubbing, because of SEO considerations as well as disability accessibility.

Social Media Videos

Most consumer and SMB facing companies have a social media presence and use social media video to promote themselves. Because social media is more targeted and conversational translating these assets can get in the way. If you have regional sales and marketing teams, it is best if they own this and create videos that are targeted to the region, feature people representative of the region, and discuss relevant use cases.

Product Demos

Product demos, especially for an important feature release, are a special case where you may be better served by shooting videos for each major language. This is especially true if you are displaying a live demo of the feature because it may look off to watch a English product or feature with foreign language subtitles or voice over. This is more expensive to reshoot, of course, but for important feature release videos, it makes the difference between a video that is translated and a video that is tailored to the audience.

Differentiating By Visibility Level

If you have a large video catalog, one of the things you can do is use visibility and usage metrics to automatically decide the level of human effort to be applied to asset captions and subtitles.

Tagging

If you know a video is going to be high visibility, tag it as such, so that the import agent routes it to a high touch workflow, regardless of the visibility metrics.

Auto-Routing Based On Visibility

For assets that aren’t explicitly tagged, you can use the following heuristic:

Low visibility (P3) >> Send to a fully automated workflow
Medium visibility (P2) >> Send to AI + human workflow
High visibility (P1) >> Send to human only workflow

The general rule of thumb is to use AI first, and then if an asset rises above a defined threshold, you then send it to human review. This is a good way to optimize for speed, cost and quality (just as we recommend for other content types).

Conclusions

If your video catalog is primarily instructional content hosted on Vimeo or Brightcove, CaptionHub is worth evaluating. If YouTube is your primary channel, plan for the workaround described above. If your catalog is primarily social media short-form, let regional teams own it.