Leveraging Cloud Technologies in Subtitling – Task Management, Artificial Intelligence, Security
PFT Blog Team | 06 Apr 2018

Leveraging Cloud Technologies in Subtitling – Task Management, Artificial Intelligence, Security
Leveraging Cloud Technologies in Subtitling – Task Management, Artificial Intelligence,… Click To Tweet

By Animesh Swain, Product Manager, Product Line Management

Over the past few years, videos have out-burgeoned more traditional modes of communication and entertainment such as text, static images and audio. Videos are so readily available across screens that absence of supporting elements like subtitles are now seen as an inconvenience, especially if the viewer is hard of hearing, is in a noise-sensitive environment or simply if the content is in a language the viewer doesn’t understand. Subtitles have already become a legal mandate in several countries. For businesses, subtitles help increase reach, view time and contribute to SEO.

Creation of subtitles has conventionally been a manually intensive task with desktop based applications such as Subtitle Edit, Subtitle Workshop, EZTitles and Cavena being the most widely used tools. With globalization, as businesses target newer geographies, subtitles for an asset need to be created in more languages than ever before. Sourcing skilled resources for different genres, languages or tapping into freelancers’ pool of talent becomes an important requirement. There is also a pressing need for businesses to reduce TAT in order to free up resources to cater to the sudden surge of volume of videos that need subtitling as well as to take up rush hour jobs where content needs to be subtitled just hours before going on air.

This makes scaling up subtitle operations a challenge for several organizations.

There is also an ever-increasing need for security. About 40% of the leaks happen during the localization phase. Especially for organizations that deal with pre-release content, there is an additional overhead to take adequate security measures to prevent counterfeiting. For those looking to scale and diversify, this translates to significant CapEx and OpEx.

Smart cloud based solutions overcome several shortcomings of conventional methods and can help businesses scale with minimal up-front costs. Broadly, benefits from smart cloud-based solutions fall under three categories - Smart Task Management, Subtitling on the Cloud and Multi-Layered Security.

Smart Task Management

Based on deliverables, organizations split subtitling workflow into sub-tasks followed by multiple quality checks. It is common for each sub-task and the quality check processes to be further divided amongst multiple professionals. The professionals may be part of an in-house department, freelancers or from a third-party vendor or a mix of all these. Much of these processes are handled today in an inefficient manner on an ad-hoc basis.

Challenges include:

  • On-boarding resources for multiple languages / genres
  • Scouting the right freelancers
  • Managing task allocation and task monitoring – most of it currently handled through spreadsheets, calls and mails

A smart cloud-based MAM with an integrated task management module can effectively handle these complexities by auto-assigning tasks to resources based on key parameters like skillset, availability, cost and ratings, thus freeing up managerial time. The solution can also auto-execute business processes on detecting completion of tasks assigned. The manager can view progress of each asset as it goes through the workflow in a single dashboard instead of endless calls and mails. Machine learning algorithms and analytics can be used to aid decision making process by analysing metrics like asset-types, user-preference, deliverables, time spent on QC etc. to determine success rate of auto-assignments and task executions and explicitly suggest cases where a manual override is recommended.

This eliminates the need for tedious spreadsheets and multiple follow-ups via calls and mails. Automating these tedious processes mean professionals spend most of their time where they add most value – in turn increasing efficiency with lower TAT for the subtitling workflow.

Subtitling on the Cloud

Technical advances in Artificial Intelligence hold the potential to change the way businesses execute subtitling operations. Early adopters in the media and entertainment industry have looked into speech-to-text and machine translation to automate creation of transcripts and subtitles. A subtitling tool empowered with AI can pass qualified genres of content to an AI engine to generate subtitles that can be straight away taken forward for quality checks. The tool in such cases needs to be designed to enable faster QC:

  • Highlighting points of low accuracy by AI by a heat map of confidence scores and audio graphs.
  • Elaborate keyboard shortcuts for most QC task
  • Programmable subtitling settings that auto-assign to specific content types
  • Ability to auto-fix subtitling errors

Subtitling on a SaaS platform also means businesses pay for what they use, thus reducing upfront costs and focusing on OpEx as they scale.

Multi-Layered Security

For media and entertainment industry, security is paramount, especially when operations involve pre-release content. In the conventional method of subtitling using desktop-based applications, each resource in the workflow needs a copy of the asset. For a 90-minute movie, generally there are dozens of copies being downloaded to as many hard disks. This increases chances of a possible breach in security, and thus necessitates appropriate counter-measures. Infrastructure expenditure such as strong rooms, expensive security software, dedicated security teams become a must.

With subtitling done on the cloud, security needs can be met with a much lower Capex. A multi-layered approach to security ensures there are fewer chinks in the armour. The layers can be broadly classified as:

Deterrence

  • In-stream watermarking on the assets which includes name of the person logged in and dynamic time stamps.
  • Watermarks on essences such as transcripts and subtitles - both offline and online during edit

Prevention

  • Integration with robust identity management software such as OKTA
  • Encrypted assets and essences at rest
  • Efficient rights management for view / edit / downloads
  • “Just-in-time" access to assets and essences for each resource
  • Password protected offline files

Traceability

  • Forensic watermarking on assets - enables agencies to track down source of breaches in complex cases.

The Road Ahead

Cloud-based solutions are becoming advanced in ways that combat obstacles in professional video workflows. But, their biggest advantage is in the agility to adapt to the future. Developments in speech-to-text, computer vision and cloud security are happening at an exciting rate. Media management on the cloud ensures organizations are quick to reap benefits while minimizing video latency and Capex for hardware and infrastructure.

Click here and we'll get you a meeting with our Subject Matter Expert

Comments


Subscribe to get emails delivered to your inbox

Email *
FacebookTwitterLinkedIn

Thank you!
Our team will get in touch with you shortly.