BBC Subtitle Guidelines | Hacker News Reader

BBC Subtitle Guidelines

bbc.co.uk

134 pointsrogual82 comments

One of those big difference between UK subtitling and US captioning is the use of colo(u)r or the absence of it (https://www.bbc.co.uk/accessibility/forproducts/guides/subti...), which I believe boils down to technological differences between television systems in the '80s. While the UK-developed teletext system is also available in NTSC (in fact CBC have deployed it), the US opted out to develope a separate captioning system that only supports basic text and positoning.

  I'M NOT SURE IF THIS IS BECAUSE
  OF GOOD OLD PROTECTIONISM OR
  CONVERTING STATIONS TO HAVE A
  TELETEXT COMPUTER ON EVERY
  STATION IS IMPRACTICAL IN THE US
  WHERE, AT THE TIME AT LEAST,
  THERE ARE A DIVERSE NUMBER OF
  BROADCASTERS WHICH MIGHT NOT
  AFFORD THE EQUIPMENT NEEDED
  SINCE TELETEXT IS AN "ACTIVE"
  SYSTEM WHEREAS THE PBS-DEVE-
  LOPED SYSTEM CAN BE USED ON
  EXISTING, CAPTION-UNAWARE
  SYSTEMS

... while the UK only had two broadcasters at the time, which are the BBC and the IBA (Independent Broadcasting Authority*), it is easier to develop and actually broadcast teletext.

* Despite their name suggesting that it only monitors broadcasts, they are actually the broadcaster which has the authority to install teletext equipment. The then-ITV companies are only program suppliers to the IBA, not broadcasters in their own right.

That's fascinating, to color-code by character.

For (foreign-language) subtitles that seems distracting and unnecessary, since even if you don't see the character speaking, you can recognize their voice.

But for closed-captioning (for the hard of hearing), it seems like it could add clarity. Yet in the past I've watched movies with CC without sound (like on a long-distance bus) and don't remember ever having a problem understanding who a line belonged to.

Does anybody have actual experience with color-coded captions and whether they're more of a help or more of a distraction?

Why would it be distracting to know who’s line it is without color coding? Why would color be the thing that distracts you; wouldn’t the actual captions be far more distracting?

As to why: it’s another datapoint to use to reconstruct what fully able people can do quickly. Sometimes captions aren’t well timed. Sometimes a lot of things happen very quickly and it’s hard to keep track. What if the camera isn’t on anyone’s face? You can probably figure it out by context, but why expend the mental energy when it can just be colored?

Just because you never had this issue (and I’m sure you did, it’s not like we remember the most mundane of details like this years later) doesn’t mean it’s not an issue.

Foreign language subtitling and captioning for the deaf are different because their audiences have different needs.

I mean, no, it plainly isn’t.

You have a single set of guidelines (BBC doesn’t have two sets, nor does any other provider I know of) and a single set of subtitles as a result.

This is accessibility like anything else. High contrast can be used by many people on the ability spectrum for multiple various reasons. Subtitles are the same. Some use them because they’re deaf, some use them because they’re not confident enough in a language to use only one sensory input, some just like subtitles.

But since your claim is they’re separate, please show me exactly how they are separate. Show me the two separate style guides, show me providers differentiating between them and providing both captions and subtitles. Show me even the smallest sliver that backs up your claim. And, more importantly, show me the exact reasoning why color would work in one but not the other.

Captions like “[ upbeat music ]” or “[ computer beeps ]” signal to deaf listeners information about the program that they cannot sense. There is no need for those when adding subtitles to a foreign-language program made for a hearing audience.

Similarly, subtitles are necessary for on-screen foreign language text (e.g., signs or documents), but not when the text is in the language that a deaf viewer understands. For example, you wouldn’t have an English subtitle for an on-screen sign in English, but you would for an on-screen sign in Chinese.

The BBC produces content exclusively (or nearly exclusively) in English, so these guidelines are surely directed towards captioning of English-language content for hearing-impaired, English-reading audiences.

The older “608”[1] system in North America was much simpler but the current “708”[2] standard does support specifying colours and fonts, but in my experience in the industry nobody uses those functions at all and just uses the 708 function to embed the older 608 payload data within the newer 708 data structure.

In UK/Europe the older/simpler format would be OP-42/OP-47 Teletext[3] which can be used for captions instead of the full-screen data pages, or DVB Subtitles[4], which get into more uses around “subtitles” in terms of language translation, rather than only the “closed caption” use case where it matches the content language. DVB subtitles can be sent as pre-rendered bitmaps or as data for client-side rendering.

[1] https://en.wikipedia.org/wiki/EIA-608 [2] https://en.wikipedia.org/wiki/CTA-708 [3] https://en.wikipedia.org/wiki/Teletext [4] https://en.wikipedia.org/wiki/Subtitles

Yeah, I know of the "708" captioning system but it is surprisingly underutilized* by broadcasters. I think that they don't see any use for e.g. color?

* in terms of 708-only features, not on the pedantic "ATSC uses the 708 system"

There are very strong lobbying groups that push for accessibility in terms of captions (as well as the “DV” described video audio track) but my impression is that their focus is on the quantity of content that’s covered, and the quality (spelling, time-alignment), and I guess they don’t care as much about text styling.

The requirements are quite high in Canada[1] and have been expanding in the US as well[2].

The company I work for makes products for broadcast customers, around asset management, linear playout automation, and the playout servers that insert the captions (from files or live data sources) so working out how that all happens is part of every big project.

[1] https://crtc.gc.ca/eng/info_sht/b321.htm [2] https://www.fcc.gov/consumers/guides/closed-captioning-telev...

708 is generally used for HD and 608 for SD. Until we rid the world of standard definition broadcasting, 608 is here to stay.

In my experience with a few big broadcasters like Paramount (previously Viacom) and Discovery, for broadcast in Europe/UK the signal they generate often has a mix of Teletext and/or DVB inserted based on the channel since those signals are distributed to a LOT of partners like local satellite and cable companies who can decide which parts of the signal to map into their system.

In that context, “teletext” just works the same as the North American 608 captions and has nothing to do with the older full-screen data stuff. There are no restrictions around authority for “teletext equipment” - for those channels they actually use systems fully based in AWS with the playout engine running on an EC2 instance so all software-only to generate the single MPEG transport stream with video/graphics + audio + captions + subtitles.

It’s also common to send a DVB subtitle multiplex that has a number of languages (up to 20+) embedded in the same signal.

To be fair, I'm talking about the old PAL-based teletext and not about DVB teletext. Also, computers are now everywhere so the 888 page can be inserted with minimal effort unlike back then that it requires coordination with different teams.

I was talking about the older PAL-based teletext as well - but you're right, I understand it was different in terms of equipment originally with the full-screen data feeds.

Separate from DVB, the old PAL OP42/47 teletext payload gets inserted into the MPEG-TS using SMPTE 2038[1] then when decoded, it would go into regular ancillary data of the uncompressed stream per ITU-R BT.1120-7[2].

The broadcast industry really loves standards[3].

[1] https://ieeexplore.ieee.org/document/7290549

[2] https://www.freetv.com.au/wp-content/uploads/2019/08/OP-42-C...

[3] https://xkcd.com/927

I was under the impression the "teletext" style ones still used teletext encoding, not EIA-608? It's basically just a data steam containing a single page of teletext rather than a whole magazine? AFAIK Sky Digital uses this approach (at least for SD), with a more modern looking decoder, and it certainly has colour support at least[1].

[1] Although DVB can contain a full teletext stream that can be reinserted into the SD analogue VBI by the receiver. Sky boxes supported that so on some channels you could just go to ye olde page 888, although I haven't a clue if any channels still do that (I don't have an old SD Sky setup around to look).

That's correct - 608/708 is North America only, and UK/Europe could use OP42/47 teletext for simple captions (not whole magazine), and/or DVB (mostly language translation).

The Closed Captioning System for the Deaf - which became known as "608" - was always capable of displaying captions in color.* However, this faculty was largely underutilized, due to a variety of historical / technical reasons.**

* Indeed, the first test broadcasts from 1979 included color.

** This is due to the design of the original 1980 "decoder," which only 'worked well' for color when installed inside an NTSC television. For the external adapters, grayscale instead of color was displayed. The only way to view color captions in color back in 1980/1981 was with a TeleCaption TV - and there was only one model ever made.

Something I've wondered... when politicians give speeches they often have some hand gesturing from a sign language interpreter standing next to them. But I've never understood why this is better than subtitles. If you're deaf then wouldn't you rather read text than follow sign language?

I would guess that the sign language interpreter is translating in near real time in live speeches.

Live closed captions (text only) is very common and standard. Usually those are done by an external company listening in to an audio feed, and sending the data back. It used to be done with regular POTS lines and telnet, but now it’s obviously more common to use public internet based services like EEG iCap[1]

I don’t know too much about it but I had read recently that ASL sign language can be thought of as a different language, rather than a direct equivalent to text subtitles[2].

[1] https://eegent.com/icap [2] https://imanyco.com/closed-captions-and-sign-language-not-a-...

> I had read recently that ASL sign language can be thought of as a different language

Yes, it is a different language. I've heard that ASL is rather similar to French Sign Language and quite different from BSL (British Sign Language). If someone were to translate something from English into ASL, and someone else were to translate the ASL back into English, I'd expect the result to be as different from the original as if they'd gone via some other language, like Italian, for example.

There's a variety of running jokes that Italian is half sign language anyways.

(Apparently derived from the fact that in Italy, there's quite a lot more non-verbal communication with hand gestures than other parts of the world)

There's more to speaking than just the words. Sign language can convey inflection and emotion in ways that closed captions cannot.

Watch someone signing: they use their face and body to convey emotions like anger, confusion, hesitation, love, joy, and the rest of the human range.

But if you're watching the politician speak, you can already see all of the emotion in their facial expression and body movement.

And that's the "original" emotion, it's not filtered through another human being. When you watch a movie without audio and with subtitles (like in a bar or on a bus), the emotions of the speakers are already awfully clear from the visuals.

No I think you're missing the grandparent's point. Inflection in spoken language "translates" to facial expression in sign language.

No I'm getting that completely. But inflection in spoken language is redundant to a large degree with facial expression. If you're watching the original speaker, you're already getting that.

For example, if we ask a question, it's not just that our voice goes up at the end. Our eyes move in a certain way too, slightly more opened and our eyebrows and sometimes cheeks raise.

And you’re seeing this nuance while simultaneously reading subtitles? From a speaker at a distance? GP is correct, sign is a fully expressive language, far richer than subtitles.

Not who you replied to, but weren't we talking about televised speech? (We're talking about adding subtitles.) So the "at distance" really isn't an issue. And yes I can watch subtitle and speaker's face at the same time and get their expression.

Yes of course. Haven't you ever watched a movie without audio and with subtitles turned on? It's quite easy to get the nuance. People's faces are incredibly expressive.

And nobody's at a distance, the cameras are always on either a medium or close-up shot when filming politicians speaking.

This is a deeply strange line of thought to follow. Do you think broadcasters would go to the trouble and expense if there was no value for people in it?

If subtitles were equal value or even “good enough”, they’d be used exclusively. That they aren’t should tell you something, and you repeatedly protesting that you are unable to comprehend the value doesn’t mean it isn’t there.

Not everything that's spoken in a televised event has a single accompanying speaker to watch for cues the entire time.

Separately, relatively dry sarcasm can't be visually picked up directly, but a signer may be able to suggest some of that with body/hand language.

> But inflection in spoken language is redundant to a large degree with facial expression.

I don't know how you'd possibly attempt to objectively quantify that degree, but my guess is that you're understating it. The entire deaf community is probably not mistaken about which means of communication are the most effective for them.

> But if you're watching the politician speak, you can already see all of the emotion in their facial expression and body movement.

Yes, but a deaf person has trouble hearing how the speaker is speaking. They're missing out on the emotion in the voice.

Consider all of the emotion that can be conveyed in an audiobook, or on a phone call, or through music. Humans convey a lot of emotion in sound that is not represented visually. Part of sign language is conveying the emotion usually present in speech.

> But if you're watching the politician speak, you can already see all of the emotion in their facial expression and body movement.

No, you can’t; how much emotion is shown via those things vs. tone, volume, and other auditory cues varies from speaker to speaker and speech to speech; sometimes, speakers demonstrate one emotion through gestures but indicate that it is insincere/being mocked/etc. via vocal cues, even.

Not to mention the degree to which simultaneously tracking face and subtitles makes you likely to miss parts of either or both.

Since I live with someone whose first language isn't English I basically watch everything with subtitles. And live subtitling - while it's a thing - isn't that good. Try turning on subtitles for something like a news programme or live broadcast and it'll usually be quite delayed, with lots of misspellings and even outright wrong text. (This is true for premier public broadcasters in the UK such as the BBC, I don't know if this is solved better in other countries).

Anyway my theory is that sign language interpreters may be much better at this because sign language uses the same areas of the brain as speaking[1] so they're able to listen and sign much more intuitively than typing. Think if you were able to listen and speak at the same time without your "speech" drowning out what you are listening to.

[1] https://maxplanckneuroscience.org/language-is-more-than-spea...

Nobody can type fast enough on QWERTY to keep up with human speech, so the comparison is versus chorded typing on a stenography keyboard, or someone repeating the dialogue into voice dictation software which is obviously prone to errors.

As someone who watches TV news with subtitles on, whatever entry technique they're using, the result is not very good [in the UK, can't speak for other countries].

Sure, but the closed captioning is still extremely good. The typists use chorded keyboards for speed and yes they occasionally make mistakes but everything is generally quite clear and accurate.

On the other hand, signing involves actual translation, not just transcription, which is much more likely to drop meaning or introduce confusion. Translation is already hard enough, and live translation is a whole other level of difficulty.

I'm not deaf, don't know any sign language, am not close friends with anyone who does.

However, I understand that it's much easier to be expressive in sign language. Non-verbal language used by the speaker - sarcasm, tone, inflection - either translate badly, or get lost entirely when transliterating into subtitles. A talented sign-language translator is able to carry this over much better.

Yes, but.

Now, you’re getting the inflection of the interpreter and not the speaker (derivative). Certainly you’re not hearing inflection on the original orator, either, so maybe it’s a mixed bag.

Sign language (at least ASL) involves a lot of facial expression in addition to using your hands. There's a lot of room to creatively translate the unspoken message in a vocal inflection into signed message.

Subtitles in general, and live captioning especially, avoid editorializing or taking creative license- they're literal translations of words without other cues added in.

One nice exception to this has been the community subtitling work involved in various Taskmaster spinoff series, where explanations for cultural jokes or idioms are added into the subtitles for foreign audiences, especially when the joke is a pun or relies on a mispronounced word to make sense.

I don't know much about sign language. Does it use a different grammar from English?

Yes, it's best to consider it an entirely different language rather than replacements of words.

I learned UK and USA use incompatible versions of sign language from online discussion of a recent television show.

Transcription can lose meaning, as text is not a 1:1 replacement for speech. You lose cadence, stress and emotion.

>Sure, but the closed captioning is still extremely good. The typists use chorded keyboards for speed and yes they occasionally make mistakes but everything is generally quite clear and accurate.

Live subtitling (on the BBC at least) is mainly done using re-speaking and voice recognition, rather than typing.

But it's being done by professional translators, right? That might be more expensive than having a human do live subtitle transcription, but I bet it's not a huge difference especially relative to the production costs of any broadcast to a large audience, and based on the live subtitles I've seen (mostly on national sports and news broadcasts, so I don't know if that's automated or done by a human) it's hard imagine the quality achieved by professional signers wouldn't be significantly better.

We have subtitles on all the time for my little boy and can attest that the BBC is very poor - even on iPlayer.

Interesting side note: if ever subtitles are turned off, or we are watching TV elsewhere, me and my wife can't 'hear' well. Even if the volume is up. Like we've untrained our ability...

Totally agree, the "live" subtitling on BBC is remarkably bad.

It's way worse than even using a cheap computer with open source solutions - it's strange no one at the BBC decides to resolve this as it would be easy and could easily give the public a far better result. Even if you took a hit on the most obscure words, off the shelf would outperform the current process by a country mile.

Surely not, syncapatocaption is infallible.

Depends, if the speech is IRL-first and video-second then a sign language interpreter is better and cheaper than installing some sort of concoction to display live subtitles (which have to be typed by a paid steganographer).

But in any case, deaf people still have the need to practice reading their language, so removing it from everywhere except IRL conversations might be detrimental to them

Sign language is the native language for most deaf people, while subtitling is derived from spoken language which is usually their second language. Also you can express emotions better using it, similar to how it's easier to convey them using speech than using text.

The other part of the jigsaw that a lot of people don't realise at first is that sign languages are distinct languages from spoken languages. Or to put it another way: ASL is to American English as Portuguese is to Korean.

Sign language has a different grammar. At least in British Sign Language. Simplistically, put the object of the sentence first so it's clearer what's being talked about.

For someone who is profoundly deaf from birth and who can't lipread, the way we speak and write is a massive struggle. Cochlear implants before a year old are much more common now, while the brain is still more malleable, so there's maybe less and less deaf people who are totally profoundly deaf and you may not realise what it's like for them if you never come across them.

Interesting perspective. I've never considered how much of reading and writing is dependent on first listening and speaking. I guess it makes sense, since the first steps to reading are "sounding out the words."

As I understand it, BSL is a fully different language to spoken English, with different grammar and syntax. For someone whose 'first' language is BSL, reading subtitles is more like a second language, where meaning is not conveyed in the same way.

https://www.british-sign.co.uk/what-is-british-sign-language...

Think about how you learn to read - you 'sound out' words, turning letters into sounds to match them to a pronunciation to figure out what word is represented.

Now imagine trying to learn how to do that when you have never heard any words spoken out loud.

People who are deaf from birth often have a lot of difficulty with spelling and reading, because both skills are closely connected to saying and hearing words. Connecting written words to lip movements (which is kind of the closest thing to 'phonics' for a deaf person) is lossy - the letter-to-lip connections are fuzzier than letter-to-sound, and lip-to-letter is very ambiguous.

Subtitles are great for people who are confident and comfortable readers - say, people who have become deaf due to age - but for some deaf people following subtitles can be like asking someone who's dyslexic to quickly read a sentence out loud.

The more salient point is that English is going to be a second language for people who grew up deaf (with a completely unrelated sign language probably being the first language).

That doesn't sound too different from learning Mandarin or any other Chinese dialect. There are some parts of characters which are roughly phonetic, but generally speaking you're not going to know how to write a word by hearing it, or how to speak a word by reading it.

Sign languages are not coded speech. They are languages with their own grammar and vocabulary. For example, American SL is descended from Old French Sign Language and is partially understandable by French SL speakers today, while British SL is completely different, not in the same language family. It is even possible to write sign language. it is done like with spoken language. The most basic components, akin to phonemes in spoken language, are a closed set, assigning a symbol to each allows lossless transcription. Mostly used by linguists; but there are some books in ASL.

Deaf people who speak sign language natively approach English as a second language. And it is hard to learn a spoken language when deaf. English literacy rates among ASL native speakers are rather low.

This has to be a generational thing. There is no way a deaf person growing up now is not going to be using the internet.

Partial literacy is all you need for YouTube or video calls. Plenty of hearing people can read well enough for that too, or to find what they need at the store, but can't read well enough to e.g. summarize the main points of a newspaper story. I've seen estimates that something like 20 - 40% of Americans are functionally illiterate in that way. For the Deaf, it is even higher.

A speech and language therapist tells me that spelling and reading is hard for deaf children because we match phonics to text but they don't have access to phonics, so they have much lower literacy levels without specialist help

Still, I suppose that many people still want the original phrasing, not a translation where subtleties might get lost.

Closed captions are usually available for that need (which is also helpful for people who became hard of hearing later in life). But that's a separate need than what is effectively translation.

Seems as relevant a place as any to highlight the lack of support for subtitles in BBC iPlayer on AppleTV: https://www.bbc.co.uk/iplayer/help/questions/accessibility/a....

I work in a related industry so fully understand how these situations come to pass, but as a user it’s frustrating to say the least.

Is it only Apple TV (and not on iPhone for example)? If it's Apple-wide I can imagine it but I'm curious what is the possible reason that BBC can't support having a separate subtitle renderer?

I just checked, iPad has it (and using colour coding, as that has come up elsewhere in these threads). On Apple TV a subtitle menu appears in the time bar, but it doesn't seem to do anything in either the "Automatic" or "CC" options it gives?

CC is a bit of an Americanism, perhaps it is running more directly though some Apple playback code that is more particular about subtitle formats?

I suspect the ultimate case is probably low usage statistics not making it a priority. The icon / BBC logo didn't update at the same time as the iOS version, so I suspect its a separate codebase for some reason?

iPlayer on TV (across all platforms) is a generic web application, topped with a custom wrapper app (think webview) for each platform, responsible to hook platform’s native APIs to Web/JS APIs.

I’m guessing there are some complications hooking Apple TV’s native subtitles APIs to relevant web APIs, and low usage statistics doesn’t help prioritising fixing the issue. Although the rumour is that they are working on a completely new Apple TV app.

For most platforms this is true, but not Apple TV. I don't think they even allow it. iPlayer is definitely native code.

I think Apple must allow it, either that or Google have tried really hard to capture in native code the "shitty non-native web app" feel with the YouTube app. But the iPlayer app does feel fairly native.

Apple TV and AirPlay doesn't support "out-of-band" of subtitles into playback or the use of a separate subtitle renderer, the subtitles have to be linked in the HLS manifest. For a lot of OTT platforms, subtitles are processed and handled separately from AV media, so this is difficult to fix. Some platforms work around this by doing manifest manipulation either server or device side, both have pitfalls.

Only recently the UK government style guide was on the front page:

A simple guide on words to avoid in government - https://news.ycombinator.com/item?id=34104530

Kind of related (to broadcast TV, not online media) - broadcasters are required to subtitle British TV in most cases. Tom Scott has a video about it: https://www.youtube.com/watch?v=m__OZ3ZsO4Y

Man, a lot more goes into syncing the subtitles than I thought.

I recently tried using bazarr to auto-sync subtitles, thinking that for some reason the autosync would improve things, even though I usually didn't have any problems. Well, after that, all my subtitles were horribly out of sync. FFsubsync did not work as magically as I expected. I ended up just resetting them all to their original versions and disabling automatic subtitle synchronization, which I will only attempt to use if the subs are out of sync for some reason.

Note that I did not use ffsubsync directly, but IIRC that is what bazarr uses behind the scenes when "sync subtitles based on audio file instead of embedded subtitles" is checked.

I didn't think about subtitles as a default for people who are hard of hearing. I thought most people would have a device that turns the captions on and off.

The reason I even noticed this with BBC is because I was interviewed by Reuters a few years ago and the footage was spread through news throughout the world both online and on broadcast TV.

A friend from the UK sent me a screenshot where the BBC had captioned me. They were the only channel to do this (that I saw or heard about) I'm Canadian, and we tend to have one of the easiest to understand clearly spoken accents. So it surprised me that BBC went to the trouble of adding captions.

This makes sense now.

I recently went through Netflix's subtitle guidelines when I was trying to figure out how to alter a subtitle file that I'd found that was pretty barebones. One of the interesting things about it was reading a rule, and then understanding the justification associated with the decision. Some of it is relatively arbitrary, but after you give it a bit of thought it makes sense.

None of it makes the slightest bit of sense.

Netflix subtitles are HDR maximum white. On a modern TV while watching dark scenes like in the House of Dragons, it's like trying to look into a cave while someone is shining an LED torch in your face.

Oh, and they won't let you chose arbitrary subtitle languages.

Someone at Netflix wrote the code -- it wasn't an accident! -- to deliberately restrict the available languages to English for the the Hearing Impaired plus four of the most common local languages.

Not speaking one of those four? Tough titties.

Would you like to see English subtitles without "Bang!", "Loud breathing", and "Somber music"?

Haha.. no.

Someone made decisions at Netflix that made sense to them, and no one else.

Do they have a guideline for WHEN subtitles are used?

I remember sitting in a hostel common room the day after Mark Duggan was murdered[1] and half joking "Are they giving that Scottish guy subtitles as a joke?" and being told "This is a very serious situation".

The previous night, the "mates" had been "joshing" me about how much more violent America is... then their eyes went wide when I was like "Well, that's a BRITISH cop car on fire, I'm switching to water... it's gonna be fun finding our way home... doesn't the queen live near here?"

(To be fair, the "chavs" were very respectful once they heard an American accent and let us walk home, but part of that may have been because they didn't grok that they check us for guns at the airport and that rumors every bald black clad hacker is in the CIA are greatly exaggerated so they totally could have knifed us all and taken our iPods, lolllll.)

[1] https://en.wikipedia.org/wiki/Shooting_of_Mark_Duggan#Shooti...

Are you ok?

Ooh, nice. I especially appreciate seeing jokes mentioned in there as a specific case. I often watch with subtitles on and it's frustrating when a joke is spoiled by bad subtitle timing.

Couldn't find anything about the use of (sic) or [sic]. Maybe they want leeway on that so as not to be offensive?

The default position is that subtitles are a verbatim transcription of what was actually said, so there's no need for "[sic]".

Reply on news.ycombinator.com