Speaker 1: Closed captioning is the process of displaying text on a video screen to provide transcription of the audio portion of a program. When captioning is applied in advance to pre-recorded programming and movies, there is time to perfect the text and match the timing of the captions to the program's speech. However, live programs like news and sports require the captioner to generate the captions as they listen to the live broadcast. As a result, live captions may include errors and may also fall behind the spoken word, even when prepared by highly skilled and trained captioners. In this video, we'll show you the two forms of live captioning. Steno captioning, where a captioner uses a keyboard to transcribe the audio onto the screen and voice writing, where the captioner repeats the spoken audio of the program and voice recognition software transcribes it. Let's begin with Steno captioning.

Speaker 2: My name is Sandra Burns. I'm the Director of Training, Recruitment and Development for National Captioning Canada. Live captioning is used for anything that is live programming, so news, weather, any sports games, political debates.

Speaker 1: Sandra works from home and connects remotely to the TV studio. Each captioner researches the content of the television shows they work on. With news programs, for example, they scour websites of major news organizations, learning how to properly spell and pronounce the names of the people and places making headlines around the world.

Speaker 2: I will input all the relevant data into a database, which is very particular to how I write.

Speaker 1: Steno captioners use specialized software that translates the words into a kind of shorthand, a special language that allows the stenographer to produce up to 350 words per minute.

Speaker 2: It takes a lot of focus to caption. You literally have to stay engaged for the entire programming, and fatigue definitely sets in. There's a lot of wear and tear on your hands. It is a physical, physical process as well as a mental process.

Speaker 1: People talking quickly and over top of each other is common on television. This can make it very difficult for the captioner to capture every word accurately, and unexpected names may present spelling challenges. Due to the speed of on-air conversation during live events and the rate at which the viewer can read on-screen captioning, the steno captioner will occasionally need to paraphrase dialogue. Inevitably, the captions appear on the screen several seconds behind the action.

Speaker 2: As soon as I start writing on my stenograph machine, captions will start generating. So there is a bit of a lag. I'm hearing it. I'm writing it. It's processing through the modem. It's going across to the network and going through an encoder at their end as well. And then it comes across the television screen. We experience lags anywhere from two lines up to as much as six or seven lines. We have what we call buffer times, so that does contribute to the lag. There's probably about maybe a three-second buffer that I have to correct a spelling before it goes out on the air.

Speaker 1: Although steno captioning accounts for the majority of live captioning in Canada, the other popular method is voice writing. Voice writing is different from the computer-generated captions using voice recognition on some social media sites. Because a person controls the process and researches the program, it is much more accurate than the computer alone.

Speaker 3: My name is Josh Summers. I am a manager within the voice writing department at National Captioning Canada. Voice writing is the process by which we create captions for television and film using the human voice. Voice writing, in theory, is suitable for any type of programming. However, there are certain types of programming that voice writing is more suited to, live sports being one, newscasts also, parliamentary proceedings, faster programming, panel shows, talk shows, are less suitable for voice writing purely because of the speed of speech. Each of our voice writers will work out of a soundproofed studio. They are re-speaking everything as close to verbatim as possible that they are hearing through their headphones, and when verbatim is not possible, then they are re-speaking and paraphrasing without losing any meaning from the original dialogue. The biggest challenge for a voice writer is coping with somebody who speaks very quickly. In voice writing, every voice writer has a top speed, if you like, at which they are able to re-speak, beyond which accuracy will suffer. With voice recognition software, there is a buffer, essentially, whereby the software holds back a certain amount of speech input to gather context so that it can output the most accurate captions possible.

Speaker 1: This video was produced by Accessible Media Incorporated at the initiative of the Canadian Association of Broadcasters and the CBC. Captioning for this video was prepared by Closed Caption Services Incorporated.