multimedia localization technology Part One
The multimedia industry’s technology solutions
Driven by a massive surge in demand from Netflix, Hulu, Amazon Prime, and their competitors around the world, software tools that automate subtitling and dubbing spring up everywhere. Media localization tools – though not yet as advanced as their CAT-tool counterparts in translation – port over features at an aggressive rate and leapfrog years in development. We’ve identified more than 30 tools in this space. To begin, we encourage you to review our previous article on dubbing and subtitling tools. In the new 2-parts overview, you’ll learn the industry’s best practices on how this market is:
- accelerating localization with embedded video players
- reaching the world of voice talents and dubbing directors
- scaling up operations with distributed teams
- enabling flexibility with client review and vendor collaboration
- ensuring process visibility
Part 1. What makes a good AVT tool?
Focus on transcreation: embedded video players
According to Prime Multimedia (an AV localization and technology company), one of the challenges in media localization is that translations are sometimes not adapted for production. Not only might the text turn out to be too long but it might even be read as unnatural, containing gibberish. In Prime Multmedia’s experience, sending MT for Voice Over happens even with big titles.
Constantino de Miguel, Spanish CEO from Prime Group (France, Spain, Ecuador), believes that there exists some lack of understanding from LSPs about the whole multimedia translation process. Ideally, it must include a reference for translators – they need to watch the pictures and be mindful of the length. They then need to transcreate and fit their creation to the following production steps. However, quite often they do neither of these steps. As a result, you have to speed up the voice (which is not good for commercials, advertisements, and other multimedia content) or re-translate and rewrite the whole thing.
To nail these issues and allow “editing within the context”, audiovisual companies have had to create their own AV tools which feature an embedded video player. This functionality is also available in CATs such as MemoQ, Smartcat, and Wordbee. MemoQ and Wordbee support .SRT extension. Omniscien’s subtitling tool, Media Studio, supports also TTML (the format that Netflix uses). The latter allows for manipulating subtitle colors, position on screen, and formatting.
With dozens of solutions on the market, it’s hard to find the one with adjustable workflow suitable for your particular process understanding. For “dubbing in the cloud”, the workflow may include:
- recording (live direction of dubbing sessions and real-time collaborative script review)
- editing and mixing
- assignment of rights
For subtitle watermarking, encryptions, and speech-to-text, MT processes may be included.
While having a customizable and scalable workflow in a tool just makes sense, some solutions only provide the production part. Some offer a workflow management platform as a separate service (e.g., ZOOcore by ZOO Digital). Others (OOONA) feature capabilities for project management and invoicing for subtitling and dubbing. Companies may use these in addition to their own editing tools.
Image Source: ZOO
Various proficiency levels
The cinema and TV quality demands a certain level of functionality depth and licencing, but some AV companies don’t. Instead, a workflow may require a large number of training or marketing materials to process. Making the tool flexible and adaptable broadens the partner and client base.
As per Wayne Garb, CEO and Co-Founder at OOONA, their idea was that online tools need to be easy to learn and not complicated by unnecessary functionality. That is why at OOONA they have designed a professional solution for online subtitle creation software tools which is made to perform specifically the task you require. And the working files are genuine, right from the original vendors.
Another example of differentiating the proficiency levels is an educational license. Valeria Cervetti, a senior subtitling, dubbing, and adaptation consultant, and the head teacher of the Audio-Visual Translation courses at the European School of Translation, was looking for a subtitling tool for training purposes. In November 2018 she chose SubtitleNEXT by PBT EU. The resulting partnership introduced young professionals to the world of subtitling. SubtitleNEXT educational licence is fully functional except for imports/exports limited to 20-25 subtitles.
Following up on workflow and proficiency, if measured, these factors could result in scalable pricing. This will make tools that provide hybrid solutions for subtitling and dubbing more affordable.
Current pricing may be labyrinthian regardless of the tool. Price per seat per month depends on what components are included in the solution (for instance, translation, review, conversion, transcription, burning and encoding, etc.) Buying a full package is usually cheaper. On average, solutions with a good toolkit sell to LSPs anywhere from USD 10K to 30K. Value of a transaction at the enterprise level may get up to USD 100K.
Cloud platforms boast keeping content secure, accessible 24/7, and easily traceable within the cloud. But if you’re working with something displayable, it can be captured, stolen, and distributed despite the measures taken and NDAs signed. Without making a system extremely difficult to use, there’s little you can do from a system point of view. Because of this, an idea has emerged to reinvent “higher levels of content security.” That is why:
Image Source: ZOO
- Omniscien’s Media Studio environment is run in a private cloud owned by the end users (all the data resides in their system and only for a very short period of time comes to the solution provider). Editors work in a browser – you don’t need to send them a .ttml file or video file as everything is streamed. All the data transmissions are encrypted.
Machine learning components
Wouldn’t it be nice to get a subtitle draft directly out of a raw video in the target language? Some of the tools already have this feature available, while others don’t yet support automatic speech recognition of the source. When utilizing machine learning components, it becomes possible to not only do audio transcription and dialog extraction, but also adjust timings and much more.
According to Constantino de Miguel, modern multimedia l10n companies should bundle together to be more efficient. To that end, everyone is welcome to try Prime Multimedia’s Audio Video Manager
Image Source: Prime Multimedia
An idea of a community fellowship lies in the PBT EU’s NEXTclub, a growing SubtitleNEXT community for sharing tips and tricks where members can contribute to forums and have access to industry insights.
Some collaborative solutions allow a client to
- log in to the system and chat with the translator/actor (which adds value to the final product)
- join in live dubbing sessions to ensure visibility into the actual project stage.
To be continued
While working on this article, we talked to actual subtitling experts: translators, editors, representatives of LSPs, and recording studios. The survey showed that people are sitting on old solutions. Sample answers:
- “I use Subtitle Edit. For translation and layout it’s enough”
- “CaptionHub is very convenient”
At Nimdzi, we ourselves record Infodrops using subtitling tech which is built in to Facebook and Linkedin. Twitter doesn’t have this capability, so we use https://www.rev.com/ and https://www.movavi.com. For our purposes, basic functionality of these solutions suffices.
But there certainly exist much more advanced technologies on the market, suitable for theatrical quality and high-level users. Read our upcoming article to discover 7 examples of these amazing tech solutions!
Stay up to date as Nimdzi publishes new insights. We will keep you posted as each new report is published so that you are sure not to miss anything.