Subtitling, Workflow The Million-Dollar Question in Subtitling: insource or outsource? Charlotte CoppejansFebruary 3, 2022 Because the cost of outsourcing may be prohibitive when producing video at scale, many video publishers and broadcasters are looking at creating subtitles internally. Using Artificial Intelligence (AI), they want to reduce the turnaround time and the cost. However, when not implemented correctly, they may find themselves delivering sub-standard computer generated captions. The key to reduce the cycle time and the cost without compromising on quality, is to set up the right workflow using an ‘expert in the loop’. Whether you are using internal or external resources to achieve that, is barely a matter of taste. When it comes to accessibility and SEO, using subtitles or closed captions on video is an absolute must. High quality subtitles are instrumental to reach the widest possible audience. Last but not least, providing subtitles is a legal requirement in many countries. A study by Netflix found that more than 80% of members used subtitles or closed captions at least once a month on the streaming platform The challenge Subtitling isn’t easy. The best subtitles subtly translate oral to written language that avoids distraction. Bad subtitles are computer generated and disrupt the user experience. Good subtitles are almost invisible. Target reading speed is 12 characters per second, less than speech. Subtitles use less words. Numbers, expressions and dates are factored in. Spotting and styling may differ per country or operator. Using tools like Limecraft, you can easily create subtitles yourself. Whether this is cost effective or time-efficient depends on your copywriting skills and proficiency with the software. The dilemma With budgetary constraints and shorter deadlines, it is tempting to use a cheap and automated solution. More often than not, this is a false economy. Because you know substandard quality of the output hampers the user experience, you may find yourself spending more money and time in post-editing the results. If you let go automation and opt for a conventional approach using subtitle editors such as EZ titles, Subtitle Edit or Adobe Premiere, you’ll notice that it is very time consuming. You’re often looking at 8 to 12 minutes of work per minute of content, per target language. The larger the volume, the more likely you will want to engage with professional subtitlers, or outsourcing the job to specialist companies. There are several great generalist and specialist companies out there, including Iyuno Media Group, VSI, True Subtitles, Lingsoft Language Services, Haymillian, and The Subtitling Company. Obviously quality has a price, especially if you need support for minority languages or other niche requirements. The answer To help you deciding on the best possible approach, here is a list of checkpoints and criteria to consider. Outsourcing: Creativity – the more creativity is required to turn spoken language into written, the more likely you want to rely on a professional linguist who’s career is based around those skills. Quality – higher standards will be more easily achieved by professionals and their reputation depends on them. Turnaround time – although outsourcing requires organisation, planning and usually takes more time, when volumes rise significantly, there might be a tipping point whereby a specialised company delivers the result faster. The same argument can be used in terms of capacity. Translation – DIY or insourcing assumes fluency in the language. If you need to translate in other languages, you definitely want native speakers. Flexibility – you’ll pay a specialised subtitling production company by the minute, so you can easily spin up or wind down changing capacity requirements Insourcing: Simplicity – where you have a combination of clean audio, good speech, reasonable volume and known languages, insourcing is probably the best option. Cost savings – we have seen an evolution to insourcing subtitling. The more AI based services become, the more companies of different sizes will consider insourcing. Using a modern AI-based system to automate 80% of the work and by adopting an expert in the middle workflow for proofreading and fine-tuning, companies realise 80%-90% cost savings. Short form video asks for a shorter cycle time – when producing short-form video (1-3 minute clips published on social media), time to delivery is key. Assuming the language and the sound are good enough, AI will deliver a decent result and post-editing overhead will be low. The overall result will be available in minutes. Conclusion Media businesses and the creative economy in general recognise that high-quality subtitles drive exposure and growth of their business. To support the core of the your message, whether you are a producer, a content agency, a broadcasters or a language professionals, accurate subtitles that capture the essence of the message are key. By taking into account the points above, and by balancing the required turnaround and the workload for adjacent tasks, you can make a conscious decision whether to insource or to outsource. In either case, regardless of your preferred approach, the use of AI is important to reduce the turn around time and the cost. At the same time, it is critical to the success of your business that AI is properly trained and implemented. Hence we advise to raise these subjects as part of your employment or procurement processes. Not sure yet? Why don’t you give it a try: https://platform.limecraft.com/#auth/signup