ENCO enCaption. Full automated subtitling.

Although we are by now used to performing tests and laboratory analyses on novel products, HDR processors, new 4K cameras or IP-based systems, mundane issues such as subtitling still give headaches to many of us. This is where ENCO comes to our aid with a powerful, versatile automatic subtitling system.

 

Subtitling of live programs has always been a problem for any broadcaster around the globe. The cost of doing it manually and the experience and quality of operators were crucial for achieving a good standard and accuracy in subtitling. As soon as the first automatic subtitling systems appeared, they were swiftly adopted due to all the problems they claimed to solve, but their inaccuracy and integration problems within the broadcast chain put a stop on their mass implementation.

 

ENCO has always been a reference in these kinds of systems and its new enCaption 4 includes really complete features and quite high accuracy in operation. Let’s see this in more detail.

 

All inputs and some more

The automatic subtitling unit we tested, a one-rack unit (although a more powerful version formed by three rack units is also available), features one Blackmagic Decklink video capture card with HDMI and SDI inputs. Both SD and HD signals in multiple resolutions and standards are supported, so the video input signal will not be an issue when it comes to integrating the equipment within our broadcast chain.

 

A welcome surprise we find here is the inclusion of Newtek’s NDI standard as video signal format on IP supported. We have not been able to test this feature as we did not have a valid NDI signal, but this implementation promises an easy integration with IP environments. What is not so convincing to me is how a traditional subtitling signal such as the one generated by the ENCO enCaption unit fits within an IP-based workflow, as more simple and user-friendly solutions are available for the task of including subtitles in IP signals, although an automated subtitling solution is always interesting regardless of the standard being used.

 

In addition to video inputs, either base band or IP, enCaption is capable of reading video files and creating the relevant resulting subtitle file. This is the so-called offline model, which is very powerful for documentary bases and deep filing systems. Transcription speed in offline mode exceeds 10 times real time, which is really fast.

 

And now, what should I do with my subtitles?

The software automatically and effortlessly detects the subtitles as soon as a video signal is inserted. However, it is true that in specific environments, some words may not be found in enCaption’s dictionary. In this event, the system allows for manual insertion of said words into a dictionary, thus optimizing voice reading and improving subtitle output.

 

It is worth noting that enCaption supports more than 25 languages. This is something truly remarkable as, in addition to this, the system is capable of distinguishing between different voice tones and automatically assigns different colours for different voices. By combining enCaption with enTranslate (ENCO’s software capable of translating subtitles into as many languages), subtitling capabilities available by using only these two systems are impressive.

 

 

 

Once subtitles have been read in any of these 25 languages available, we have at our disposal several options for further integration within the production chain. If the signal will be broadcast live, we then have serial RS422 connections for linking enCaption with our decoder and base band video subtitles inserter. This is the most widespread standard, but also the oldest and most limited one. Another option is an IP Telnet connection, with is also widely used in systems of this kind and makes good use of the network infrastructure without compromising reliability.

 

Last, we have the MOS protocol for NRCS. This is a standard protocol in the industry for exchange of information, events and step outlines amongst news systems, which also handles subtitling information. My recommendation is using IP Telnet or MOS if possible, as these are the most versatile systems and use the existing infrastructure, thus avoiding dedicated wiring for subtitle insertion.

 

If, instead of going forward with our broadcast, we would like to save these subtitles into a file for storage, enCaption has export capabilities for most subtitle file formats. This feature, in combination with the offline operation option, subtitling of video files, offers us a very powerful and fast file subtitling system.

 

Streamlining the process

In order to get enCaption work properly, a number of functions are provided in the system that we must explore in order to improve the subtitles to provide, regardless of output format we decide to use.

 

Word library

Although enCaption has a huge vocabulary database for each of the 25 languages supported, some words may not be included by default. This is why there is the possibility of creating our own word library –or dictionary- for any words we may spot that the system is not able to automatically recognize. I find this option particularly helpful when it comes to including words from another language in a specific language library.

 

Filters

As opposed to a word library or dictionary, there is also the possibility of prevent the system from recognizing certain words by using prior filters. In this instance, usage is somewhat more complex, although this could be useful in regard to names of people or places that could be confused with words from other languages, for example.

 

GPI

As it is the case with other traditional broadcast systems, enCaption comes with GPIs as well. This feature is provided through an USB adaptor that transforms a serial USB port in order to have pins from physical GPIs. Basic operation functions such as start, stop, mute and the like are the only ones available. Not very advanced, but also useful.

 

Conclusions

I have been always reluctant to use automated systems for critical operations such as this one. Having said this, enCaption has surprised me positively. Both voice recognition achieved and features implemented comfortably deliver what it promises and, in some instances, with a positively surprising result.

 

Subtitles would seem a somewhat outdated requirement, but both due to broadcast regulations and to accessibility and functionality, it is something that must be offered in all our broadcasts, either traditional or digital, OTT or VOD.

 

If enCaption were capable of inserting subtitles into a base band video output, it would be the perfect equipment for inclusion as subtitling processor in or broadcast chain, thus saving a lot of time and costs in multi-language subtitling, which is always expensive and time-consuming.

 

Its file subtitling capability is really powerful and can be a highly valuable tool for deep file transcriptions in a very efficient manner.

 

By Jeray Alfageme