Blog

Live Closed Captioning Blunders Explained

The captions are horrible!  Why can’t they spell?  These are the valid questions asked when watching live closed captioning of a TV broadcast.  Let’s briefly look at the behind the scenes of live closed captioning.  Live closed captioning, in the majority of cases, is completed by a human.  Live captioners, called writers, write to a steno machine (keys are all based on sounds).  The steno machine is connected to live captioning software that has an extensive dictionary to take the steno strokes and convert to English words.  A dictionary is not just imported from Webster’s to the software but rather created per writer based on their writing theory.  So how do they know the difference between words that sound the same such as to, two and too?  They have to remember the stroke they gave that word when creating their dictionary.  So when writing 200-300 words per minute the human behind the machine can mistroke the word and you could see errors such as:

 prays Hymn when it should be praise Him

mall mighty God  when is should be almighty God

top or pop

 There are other times when a word may just clearly not be in their dictionary and it would show up as:

 graph fight when it should be graphite

authentic cater when it should be authenticator

you know sex when it should be unisex

ear gates when it should be irrigates

Aberdeen’s live Abercap department strives to write every show with a 98% or better accuracy rate.  This is completely possible as long as the following is provided:

 1)      Show information such as key words, hosts, guests, websites and phone numbers

2)      Clear audio feed to the live captioner so they can hear the audio they are to write well 

98% accuracy means you may see 4 errors per minute when the speaker is speaking at 200 words per minute. 

 To see a live demo of a live writer, watch this clip: http://www.youtube.com/watch?v=dCPEjz–BvE

If you are interested in learning more about our live captioning department or any other of our broadcast services, feel free to send us a note here.

Perfecting the Art of Voice Dubbing: The Director’s Role

Many of us in the U.S. who watch foreign films enjoy the subtitle text on the screen as a way to understand the film, however, many films or other types of programming also include dubbing, which is where a different language is recorded and is either played over or instead of the original actor’s speech. That way, the viewer can understand the film without having to read the subtitle text. This method is preferred over subtitles in many countries and by many viewers. Our clients are often dubbing their programming that is originally in English to many other lagnuages to gain a broader, global audience.

I am often asked about who is involved to make the whole voice dubbing process happen and many questions arise in regard to what role the director plays. The voice-over dubbing team usually consists of a director, talent, engineer, producer, and oftentimes, the client.  Before the session, the native director usually goes through the material and marks places on the script that could be an issue and points them out before recording. The producer is briefed on the project and expectations and he “runs” the session with the director. The talent is the person who is performing the voice-over. The director is in charge of the script, makes sure that the brief is clear to the talent and engineer, and translates the interaction between producer and/or client and the talent, in case the talent doesn’t speak or understand English. The director makes sure that the talent reads properly, with correct intonation, pronunciation, articulation of certain words and proper names, and maintains the style of the original language. The director will also suggest copy changes, correct the talent if they make a mistake, and make suggestions on how to re-read a section in case the producer or client wants a different read. The director is there to guarantee quality of the voice-over, as he or she is familiar with the original language. The group works together as a team to produce the high-quality dubbing whether that be for feature films, corporate training videos, promotional materials, educational series, etc…

Looking for voice dubbing services, captioning, or file delivery services?  Click here to send us a note as to how AberLingo, Aberdeen Broadcast Services’ Languages department,  might be able to help you.

Calling All English Majors!

When I graduated from college with a master’s in English, the question I heard endlessly was, “So, are you going to be a teacher?” For some reason, most people assume that English majors are open to only two professions: teacher or author. The reality is that English grads posses skills that are applicable to a wide range of career choices. An English degree is proof that an applicant has complete command over the English language, which translates to exceptional skills in communication, writing, and critical thinking and analysis. English grads can proofread company documents, create flawless marketing products, increase sales through social media outlets, and compose business letters, blog articles, and e-mails that will reflect professionalism and poise.

My English degree landed me a career at Aberdeen as a Caption Editor. I came in knowing absolutely nothing about closed captioning and the broadcast industry but I was eager to learn. At first the job seemed very technical, which made me a little unsure that I was a good fit. However, I soon learned that my background in English was essential. The fundamentals that I learned in college help me train transcribers on different grammar rules. My background knowledge allows me to remain confident in my ability to provide high-quality closed captioning on every one of our client’s programs. For more information on closed captioning as a career, and the positions available at Aberdeen, visit Aberdeen’s career page.

For a great article on the power of an English degree, see this article.

Taking Captions to the Community!

Last month we visited a nearby preschool during the Read Across America week, in order to introduce DCMP’s Read Captions Across America program. The students at this school are 3-5 years old and are beginner readers. Nearly all of the students watch television daily, and none of them were aware of closed captioning. We discussed how simple it is to turn the captions on and that closed captioning allows them to “read” the shows they love to watch. The teachers had never heard about the correlation between literacy and closed captioning, and were excited to know that they could turn the subtitles on in the DVD menu of certain educational programming that they had in the classroom.  The students all received a flyer that went home in their parent folders, so that the parents could learn more about the benefits of closed captioning. It was great fun to be able to spread awareness of the importance of closed captioning to people outside of just the broadcast industry. While closed captioning is intended for the deaf and hard-of-hearing community, it can be used to promote literacy and even English-language learning amongst non-native speakers.

Aberdeen plans to take caption awareness to the public on a regular basis, so stay tuned to hear about our future adventures in the community!

 

New Legislation Expands Access to Closed Captioning

On Wednesday, March 13, Senator Tom Harkin introduced two bills that will expand access to closed captioning and video description in both movie theaters and airplanes. Harkin is the Senate sponsor of the Americans with Disabilities Act (ADA). According to Harkin, “More than two decades have passed since the enactment of ADA, and in that time we have seen a transformation of our physical landscape…however we still have more to do. These bills will allow Americans with visual or hearing impairments enjoy going to the movies and watching in-flight entertainment, through captioning and video description, just as they can at home.”
The Captioning and Image Narration to Enhance Movie Accessibility (CINEMA) Act would require movie theaters to make captioning and video description available for all films at all showings. The Air Carrier Access Amendments Act would require that airlines make captioning and video description available for programming that is available in-flight for passengers.
For more information, visit this page.

At NRB: President Matt Cook Shares What’s New At Aberdeen

Save the Date! Read Captions Across America: Friday March 2, 2013

Next month marks the annual Read Captions Across America, which corresponds to the Read Across America Day, celebrated in schools nationwide. The purpose of the day is to get kids reading and DCMP promotes this day to show that educational programming can be a great tool to promote literacy if captions are turned on. On March 2, kids come to school and spend the day reading books and celebrating Dr. Seuss’s birthday. In addition, many schools are now also celebrating closed captioning by letting students watch Dr. Seuss movies while they read the captions or subtitles on the screen.  Encourage your local schools to celebrate captions too! For more information, visit: http://readcaptionsacrossamerica.org/index.html. Here you can order an RCAA kit, free of charge, which includes posters, bookmarks, certificates, and a DVD with 6 Dr. Seuss titles to show in the classroom. Spread the word!

The Continued Fight For Web Accessibility

While sites like eBay continue to grow in popularity, they are not easily accessible for all people. Melissa Earll wanted to sell items on eBay but could not do so, due to the website’s verification system which requires sellers to retrieve and submit a password from a telephone call. She can’t hear the password and says eBay doesn’t use voice-to-text or other technologies deaf people often rely upon. When Earll confronted eBay, their response was: “Can your mom or dad answer the phone for you?” Earll’s response: “I’m a 47-year-old adult woman. No. I don’t live at home. No.” The courts dismissed her case due to lack of evidence. However, many feel this case will go all the way to the Supreme Court. What do you think? How responsive must the Internet be to the Americans With Disabilities Act? To read more, visit:

http://www.cnn.com/2013/01/10/tech/web/ebay-internet-disabilities/index.html?hpt=hp_c2

Video File Delivery Simplified – Broadcaster Solution Explained

If this question is asked: How do television stations receive programming from producers? Most people outside the industry would assume that everything happens digitally file-based. We are in 2012 and this is the way things are done, right? Not exactly the case.  Moving completely file-based has proved challenging for most TV stations, where receiving tape remains commonplace.  This article seeks to explain step-by-step exactly how Aberdeen Broadcast Services has made going 100% file-based not only possible, but easier than would be expected for both producers and TV stations. You will learn about what is currently happening at stations to date, a solution, and then the steps to the process. In this article, insider information is shared from software to infrastructure used to make a complete digital file delivery to TV stations a success.

Discovering a Niche

Necessity is the mother of all invention.  This phrase holds true for Aberdeen Broadcast Services, a company based in Southern California, USA.  Originally incorporated in early 2001 as a Closed Captioning company, Aberdeen built its business around providing closed captioning services to producers and broadcasters who were forced to comply with new FCC regulations that mandated closed captioning to appear on (almost) all US television broadcasts. During the first ten years in business, the program transfer medium of choice was tape, mostly analog Beta SP tape.  Year after year thousands of tapes went in and out of the Rancho Santa Margarita, California offices via FedEx.  As the business grew, owners Matthew Cook and Becky Isaacs knew there must be a more efficient, cost-effective way to move these programs around the nation and the world.  With broadband Internet saturation reaching record levels every year, the Internet was looking increasingly like the avenue to accomplish such a task. However, initial experiences with popular file transfer services and other FTP-based options proved unreliable.  In the television business, unreliable was not going to be acceptable.

Working Backward 

The initial hurdle was moving large, high-quality broadcast master programs to Aberdeen’s offices for captioning and then getting those masters to the destination television stations for broadcast.

These large master files could be upwards of 30GB for a half-hour High Definition (HD) program. At the time, the most widely utilized distribution method of transfer over the public Internet was FTP (File Transfer Protocol).  FTP is a standard network protocol used to transfer files from one host to another host over a TCP-based network.  After robust testing, TCP was found to be far from ideal when trying to sustain a large data stream over long distances.  The biggest drawback of TCP-based file transfers is latency.  The time it takes for transferring packets to be sent from A to B and then acknowledgement confirmation sent back to the sender (B to A) to confirm the receipt is called Round-trip Time (RTT) or Round-trip Delay Time (RDT).  The TCP protocol needs this acknowledgement before the sender will send another packet.  This amount of time from delivery to confirmation is called latency; the higher the latency the slower the transfer.  The further apart the two connections are the longer the return message can take, thus causing even the fastest high speed connection’s transfers to slow to a crawl.  The longer the connection time necessary for a transfer the more vulnerable the transfer is to timeouts and service disconnects.

After looking into the broadcast transfer solutions available at the time, the software chosen was based on the adoption of the User Datagram Protocol (UDP) technology and its advantages over the more commonplace FTP protocol.   UDP is an alternate method of transfer which does not wait for packet confirmation and delivers as many packets as fast as the connection will allow.  An MD5 checksum at the end of the transfer, scans the file on the receiving end and makes sure the file is complete.  Any missing packets that may have been lost during transfer are resent to complete the file.  This improvement allowed transfers to maintain much greater data rates over optimum connection locations but offered the largest improvement for long distance transfers and slow speed connections.  Other sensible features that would assist in the logistics of file transfers such as:

  • - Eliminating connection timeouts through auto-resume features
  • - Ability of running in the background as a service on clients’ systems
  • - PC and MAC compatible
  • - A webmail function with the ability to send links to download the files manually
  • - E-mail notification
  • - Low cost per node installation.

Computer & Network Infrastructure

The decision to house the networking and computer systems off-site was made to add redundancies and protection from power outages and other service disruptions.  A SSAE 16, SOC-1 Type II data center in Dallas, Texas was chosen due to its central location in the United States and its 24/7 support.  Building out the appropriate transcoding farm was the next challenge.  The initial architecture was built on a Dell Blade chassis with a Gigabit backplane connecting a 60TB Dell MD3000 and MD1000SAN in a RAID 5 configuration.   As the digital file delivery business started to grow it was immediately apparent that the networking limitations of a chassis system connected via Windows Share (NTFS) to the SAN would not support the high resource demands of a multi-blade transcoding farm.  Bottlenecks originally limited the six blade system to a shared 1Gbps iSCSI connection.  This was not enough throughput considering the mezzanine file format for the system was uncompressed AVIs with data rates up to 600Mbps per file.  The multiple transcoding blades were quickly slowed by these large files traversing the network.  The decision was made to find a network consultant seasoned in video transcoding to redesign the system in an effort to increase throughput and scalability.  Jeff Brue at Open Drives Inc. came highly recommended from a number of West Coast post houses and major motion picture companies.  The redesign of the transcoding farm by Open Drives Inc. included a new custom 180TB Solaris Share storage system with full Active Directory support and Windows compatible ACL’s.  New SuperMicro 16 core 2.4 GHz servers with 32GB of RAM and GPU processing video cards were implemented to maximize the transcoding horsepower for each blade.  Network switches were also upgraded to support a central 10Gigabit Ethernet infrastructure.  By efficiently utilizing memory caching and 40 Gigabits of bonded 10Gb/s connectivity, a balanced mix of processing power and storage speed have been achieved.   Aberdeen operators currently remote into the ten servers at their Dallas data center using Remote Desktop and web GUIs.  The team’s operators manage, receive, transcode, and deliver hundreds of full-length TV programs and movies a week from off-site locations around the globe.

Transcoding to Match

Television stations’ method of broadcasting can easily be compared to an Apple iTunes playlist.  The machine and interface that handles this is appropriately called the Play Server.  Each track in the “playlist” is either a promo, commercial, or program and as each clip ends the next file is played and passed through the airchain to the end viewer.    Although these Play Servers can handle a multitude of different codecs, wrappers, and metadata tracks, stations usually have a “house codec” that all files are encoded/ingested to for constancy and uniformity.  In an effort to minimize the number of times a client’s program is encoded, Aberdeen transcodes the asset only once to the house codec for each station deliverable, allowing it to enter the station’s workflow seamlessly.  Important encoding  variables for individual station house codecs include codec, wrapper, bit rate, GOP structure, aspect ratio, resolution, color sampling, field dominance, frame rate, audio format, sample rate, bit depth, and caption data.

Brent Chance, the Broadcast Engineer at NBC Encompass had this to say about file delivery systems: “It is vital for digital delivery services to have the resources and means to send content in a format that is ideal for the station’s workflow, their systems demand it.”

Being a captioning company, the successful transcoding of digital files would hinge on Aberdeen’s ability to get caption data correctly into the house codecs of their clients’ stations.  This proved to be no easy task as the caption information can live in various positions inside the different digital files.  The following are different specifications for different codec/wrapper combinations for caption data to be written to.  SMPTE 436M, ATSC, DTV, SCTE20, SMPTE RDD-11, SMPTE 328, SMPTE 314, SMPTE 374, SMPTE 360, CEA-608, and CEA-708.  Believe it or not, getting correct caption insertion into the deliverable file is still an ongoing effort that involves working with the software developers that encode the files and the hardware developers that make the station Play Servers.

Automation was found to be the most advantageous way to move files through the various stages of the transcoding process.  Using the latest professional software encoding engines, Aberdeen’s engineers utilize the built-in watch-folder functionality of these programs to automatically push assets through the television station specific workflows for encoding and quality control.  Developing a folder structure that made logical sense to an operator and contained the correct layers necessary to utilize the watch folder automation was vital.

In addition to expensive broadcast encoding software, inexpensive or freeware programs have found a place in daily transcoding workflows.  At Aberdeen, programs such as Handbrake, MediaInfo, VLC, Quicktime, Windows Media Player, and JES Deinterlacer are all used daily to play, analyze, and transcode different file formats.  Most of these Freeware programs use the open source FFMPEG cross-platform solution as the engine to record, play, encode, and stream content.

Client’s Master Files

Planning for the diverse cross-section of Internet connection speeds from client locations was critical to create proper expectations and standards for client file uploads.  Most high-speed Internet, cable modem, and DSL connections are designed to allow much faster download speeds than upload speeds.  Producers need to use their slower upload speeds to submit files.  This compounded with the large file size of broadcast quality programs meant that a number of variables needed to be considered.  Aberdeen customized a popular data transfer chart  to illustrate the approximate transfer times relative to file size.  Different file compression formats were then recommended based off the connection speed and file size calculation from the chart in an attempt to minimize transfer times while at the same time not compromising the quality of the file content.

After researching codec white papers and conducting critical visual analysis of all of the commonly used compression codecs, Aberdeen approved six popular/accessible delivery formats which would be recommended based on upload speed, NLE system, and source content.  Among the codecs recommended were Apple ProRes(LT), Sony XDCAM, DVCPRO(HD, 50,25), DV, H.264, and Mpeg-2.

Receiving programs digitally not only forced Aberdeen to require their clients to refine their file specifications but also their content specifications as well.  Television broadcast standards put forth by the ATSC, DVB, ISDB, and DTMB require terrestrial, cable, satellite, and handheld content to adhere to strict signal specifications—the most notable of which are for Chroma, Luminance, Gamma, audio loudness, and peak level.  Since the station deliverable files that are being created by Aberdeen go directly to the station’s Play Server and not through the traditional tape ingest process where signals are conformed by a Processing Amplifier, they needed to devise a method to ensure each program was delivered with legal values for broadcast.  Among the many viable software options, VidChecker was chosen based on its ability to not only check for values outside of legal specifications but in most cases to correct them as well.  It is this combination of client education and the software check-correction that allows Aberdeen to process and deliver hundreds of files a week with 99.999% station acceptance.

TV Station Accepting Files

TV stations are well aware of the need and desire of programmers to deliver files digitally.  However, this can be much easier for the producer than it can be for the station.   Tapes have been the medium of choice for decades and the station engineers and technicians know how to handle them.   To many engineers, files are new and complicated, especially if stations accept a number of different formats.  There are hundreds of ways files can be encoded incorrectly and TV stations need to figure out new systems and train technicians and engineers on other ways of processing new programing.   Many stations are trying to tackle this process on their own and are realizing the complexities involved and the need for additional personnel.

Aberdeen helps TV stations simplify the file delivery process.   By creating a specific file that matches each TV station’s Play Server, Aberdeen limits the number of tasks that the operator/technician has to perform because the file is in the native format of the Play Server.   The ability to create files in station specific formats has been well received by all stations due to the reduction of time it takes to process Aberdeen files for AIR, ultimately freeing up manpower and streamlining internal processes.  An additional benefit is that the UDP delivery system being used is completely automated.  When a file is placed in the appropriate station watch folder a notification e-mail is automatically sent to the appropriate station employees stating that a new file is ready and will be queued for download at the top of the next hour.  Files are usually configured to download to a machine in Master Control or to a shared network attached storage system (NAS) for easy transfer into the station’s automation system or Play Server.  At the station’s request, the leader and tail elements (bars, tone, slate and black) of a file can be trimmed off the program to make the file prep process even more seamless.

Files are delivered to stations with a specific naming convention that allows broadcasters to tell exactly what the program, episode, and airdate are without having to embed or attach additional metadata.  The primary naming convention currently being used is ProgramIdentifier_EpisodeNumber_Airdate.  By providing an optimal file for the station’s workflow, the process of receiving outside content is made more efficient.

Some stations that started with one or two programs have been pushing all their outside producers to use file-based delivery services.  KDTN’s engineer Stephen Darsey had this to say: “I dare you to find the difference between a tape ingest and delivered file, you won’t and I don’t care how long you’ve stared at a scope and a monitor, you won’t see it!”  David Tait, Broadcast Engineer from Zoomer Media says, “(The) technology used (UDP) beats FTP and is more robust and reliable.  We also like the fact that the files are automatically delivered on a schedule.”

Processes and Procedures

Automate everything!  This has become the internal operational mantra of the AberFast business unit.  By taking as many of the manual processes as possible out of the workflows, files are run through the system and are delivered and billed quickly, efficiently, and accurately.  The manual processes that do remain are treated as opportunities for quality control and gatekeeping.  Standardizing file naming conventions for all clients has allowed Aberdeen to employ filters to automatically sort files into appropriate workflows based on the prefix characters of the file name.  Requiring clients to submit file names in the correct format eliminates manual tasks when files arrive and allows for faster and accurate processing.  For example, a file filter dialog allows operators to enter inclusive or exclusive file naming filters.  A file “JPM_KDTN_W308_092012” will go through a workflow built to include all file names with the prefix JPM_KDTN*.   This simple ideological change proved to be a very powerful tool for routing different program versions through specific workflows automatically.

Proxies are used at multiple places in the automation chain.  For example, the first task of the transcoding farm is to make a proxy in the same aspect ratio and frame rate as the asset file.  This proxy is what Aberdeen’s transcribers and caption editors use to create caption files that are later embedded into the final deliverable files.  These small proxies are usually 1/10th to 1/50 of the original asset size (usually in WMV or MOV formats) making them easy and quick to move within our network.  Proxy files are also used in the QC process.  From the final station specific file, a proxy with open/burned-in closed captions is created.  This file is then manually reviewed to ensure that closed caption data, video and audio sync, and start/end times are all accurate– similar to checking a tape.   These proxy files are available to the TV stations or producers to provide confidence and proof of captioning for the final deliverables.

Another layer to Aberdeen’s asset management workflow was standardizing the Windows folder structures of the shared storage (SAN).  This was an important practice that aided in maintaining organization throughout the shared storage array.  This consistency allows operators to easily understand where files need to be placed to start the automation and where finished files arrive after going through the job cycle.

In Summary

In this article you have read exactly how 100% digital file delivery has been achieved successfully, but not without  two years of late nights, countless meetings with software vendors, lots of trial and error, and just plain old hard work to come up with this functioning solution. We see that this process still takes extensive testing with stations, but once that initial testing is over, it is smooth sailing. Sounds like a dream come true! For the producers who use it, it is. For TV stations, it is more than a dream—it changes everything.  Get moving in the right direction and go completely file-based.

 

 

About the Author

Matt Donovan is the Director of Digital Delivery for Aberdeen. Matt is a Broadcast Engineer and graduate from Pepperdine University.   Matt has worked in many live and studio environments over  his young career  for the likes of ESPN, FOX, CBS and Current TV.  Matt is passionate about creating beautiful images.   He uses his broadcast technology experience to build and maintain Asset Management Systems that enhance the quality of Aberdeen’s clients programing.

Live translation via captions, challenging but it can be done

Aberdeen Broadcasting Services was recently contacted to present a solution for a unique challenge: providing English to Spanish translations at a live event in which a pro-football player would be interviewed in front of both an English and Spanish speaking audience. Our solution? Translation through captions.

At first glance this might not seem like a big deal, but other more conventional translating solutions wouldn’t cut it this time. A standing translator was not an option since more than half of the audience would still be English speakers. This would have also thrown off the flow of the informal interview style producers desired. On the other hand, the logistical challenges of getting radios for the several hundred Spanish speakers in the audience meant radio translation wasn’t a feasible option either.

In order to make this a success we had to make it over several hurdles. First: human resources and talent. This time we used in an in-house translator for the English to Spanish oral interpreting and we secured one of our most skilled Spanish live writers.

The second set of challenges were technical: the event would be held at a sports arena in San Diego, the translator would work out of Aberdeen’s offices in Orange County, and the writer worked out of Columbia. To solve this, the translator connected to the event through a phone line and audio coupler; the translator connected with the writer through Skype; and the writer dialed in to the encoder at the sports arena.

The third set of challenges belonged to the linguistic realm. The guest speaker was a pro football quarterback talking about his experiences as a Christian in the pro college football and the NFL; this meant his speaking would be filled with a mix of Christian and football terminology. Also; the translator was Puerto Rican; the writer Colombian; and most of the audience Mexican. This required the translator to use words and terminology that kept the accuracy, integrity and feeling of what was being said in English, including football terminology, but that at the same time would be understood by both the writer and the audience.

The last challenge was practical and related to the audience’s experience: keeping the delay to a minimum. The goal was to keep the final Spanish captions limited to a 4 to 5 second delay from the time they were originally spoken. This only gave 2-3 seconds for both translation and writing, plus the 2 to 3 second delay that is unavoidable when doing live captions.

In the end, the event was a success. All these challenges were met with unprecedented coordination and communication. And most importantly, the Spanish speaking audience’s need was met with a timely and accurate translation.

What do you think of Aberdeen’s solution to this event? Would you have done things differently? We would love to hear your feedback.

This blog was written by Rolando Betancourt