So, I was watching a show on Netflix a week or so ago on IOS, and something happened that I'd never seen before. Some of the show's dialog was in Spanish, and I was shocked to find that Voiceover was speaking the subtitles along with the dialog. For those who use Netflix a lot, this might not be anything new, but bear with me. It blew my mind, and I got a lot more excited about it than was probably necessary. After that passed though, I started wondering if a system like this could be used for audio description versus someone having to record a script alongside the movie or show. As far as I know, this isn't happening already. Since Voiceover on IOS duck’s audio by default, the speech wouldn't be difficult to hear over the show/movie's audio. On top of that, it would be faster and easier to create a system where text is queued to be spoken at certain time stamps throughout the film. Descriptions are usually already timed to take place between character dialog, so there shouldn't be much of a problem with the descriptions overlapping with that. (but I guess that would depend on how fast or slow someone has Voiceover’s speech rate set to) I'm not sure how the subtitles thing is working though. Is the text actually showing up on the screen and Voiceover is just reading it as a result (which as far as I could tell wasn't the case), or is there something more going on behind the scenes? Either way, why not use that same system to deliver audio descriptions? For those who prefer the human recorded descriptions, why not offer both as an option? (with the advantage of VO spoken descriptions being available for shows and movies that don't have prerecorded descriptions) Any thoughts?
#1 (edited by BoundTo 2019-02-10 21:04:23)
I think it's a very good idea. One of the problems with AD is not enough space in the gaps to fit much info. You could pack more info in with a good TTS voice set at a high rate, and it'd be cheaper to do than recording a voice artist.
As someone living in a country where probably less than 5% of the movies broadcast on TV is described and where cinemas only do description during very select and few specialized showings for the blind, I totally agree. There are, however, some people who really hate this idea. Netflix recently tried to do some audio description replacing a human reader with Text To speech (watch fastest car to hear how it sounds like as an example) and people complained it wasn't emotional enough. In my opinion those people are being choosing beggars.
Having description stored as a subtitle track is theoretically already possible, and such a system has several advantages - you can customize the voice, but also read the description in braille if you prefer. However, many platforms would need to be updated to take advantage of it - iOS right now is the only one which reads captions thanks to an apple API. But I think even doing what Netflix did with text to speech would be an advantage, because such a description is far cheaper to do, can quickly be translated into other languages, and doesn't have the usual package of rights restrictions associated with traditional audio description tracks.
Voiceover tries to OCR text when it can, like, during Feer loading, it said possible text, made with Unity. It will also OCR the twitch app's chat, though seemingly at random intervals.
If you have a cat, the cat owns your crib.
Sorry about that, I don't make the rules, the cats do.
For multi-cat residences, one cat owns the other cats, *and* your crib, dang, eh?
#5 (edited by flackers 2019-02-10 23:21:53)
I used to think it would be impossible to read a novel using a screen reader because it would be too flat, but once you get used to it, it's perfectly fine. Audio description is usually quite dry information, and the voice artists rarely inject much emotion into it anyway so I don't know what those people were complaining about.
VoiceOver is reading the subtitles because Apple introduced this capability in iOS 11. See Settings > General > Accessibility > VoiceOver > Verbosity > Media Descriptions.
I would not support using text-to-speech for audio description. Organisations of audio describers recommend that the people who read audio description scripts should not use the same tone in describing a high-intensity action scene and a scene of grieving. If synthetic speech were used, the audio description would sound flat and almost monotone compared to the actors' dialogue. Furthermore, sighted people are beginning to discover audio description as a means of watching movies and TV programs without looking at the screen, and they would not support this idea as they would consider the voices too robotic, and it would detract from the high-quality auditory experience. If we as blind people want more audio description, we should encourage sighted people to use it, and together we can have a stronger voice in appealing to movie studios and government bodies to increase the availability of audio description.
Braille support wasn't something that I considered, but your right. On top of that, being able to customize the voice used would be a nice addition as well. Add in not having to mess around with rights/contracts and it seems like a nice concept to consider at least. Descriptions for new films could be made available a lot sooner and without so much hassle. As far as the amount of emotion goes, that's never really where my attention is. I just want to know what’s going on that isn’t portrayed through dialog. Of course, it comes down to personal preference, but I’ll take synthetic speech for a wider variety of films over more natural speech for a limited selection. I wonder how insane it would be to compile several texts of a film's plot from different sources and use natural language processing to develop an audio description script? That wouldn't cover the timing though, plus the plot summaries would have to be very detailed and accurate. Speech recognition to detect specific lines for timing? Might be more trouble than it's worth. You can find movie scripts online at least, but it might be easier to just have someone do it themselves. Combine the two ideas somehow?
#8 (edited by flackers 2019-02-11 03:49:13)
If the majority of visually impaired people hate the idea of TTS description, fair enough,but we don't need to worry about what the sighted think of it. They may show a bit of curiosity, but movies contain so much beauty and other types of visual information that they're never going to wave that in favour of a few seconds of audio description. They want to look at the beautiful people and the spectacular cinematography. Even a scene like in Halloween for instance, where the psycho is purposefully walking knife in hand across the street towards the babysitter while she struggles to find her house keys. The blankness of his creepy mask, and how close he gets before she finally gets indoors and slams the door is so menacing, and is most effective visually. Sighted people just aren't going to willingly deprive themselves of that kind of stimulation.
I've been asking this question since voiceover started reading subtitles.
When I brought this up on applevis in the many subtitle threads people weren't too keen with the idea.
Most blind people don't like text to speech description.
I really couldn't care less, as long as we can get content described in a timely manner.
I brought up the whole subtitle description to netflix once, but I swear that chat felt like I was talking to a robot.
All they said was they will pass on my request.
There is a website, opensubtitles.org, where they have subtitles for stuff up within hours, but I guess it is a lot easier to transcribe spoken dialog.
Take a look at a service called you describe, they describe youtube videos, they even have an app.
I really don't know why something like this hasn't taken off. I mean, some of the described content have horrible mixing, with the volume turned down quite a lot just to describe something.
What we need is something like opensubtitles.org, but for description content.
I don't see this as a replacement, much like audiobooks verses ebooks, it'll just be an alternative option.
This would be very handy when watching something with family or friends because you can read from a braille desplay, assuming you can read fast enough, not to mention the deaf/blind would benefit from it as well.
I just had a listen to a bit of fastest car, it's actually not that bad, I mean, you have to put up with the TTS voice they use, which sounds like the tutorial voice they use in feer.
For me, it's tolerable.
I can't see computer-generated audio description happening any time soon. There are just too many things in a film, and audio describers have to skilfully choose what to describe. Many programs have to be fact-checked ("Who is this character?", "What is this object called?", "Where is this scene taking place?"). Movie scripts are generally written in very vague language with the director and actors deciding exactly how actors will perform the scene.
Most new content is described either immediately after the footage has been edited or just before it is released to movie cinemas or broadcast on TV, so there shouldn't be a very long wait to get the audio description.
I agree. I don't think it matters that much what sighted people think about audio description of any kind. (recorded or synthetic) Not that they don't have a right to use it, but it's not really a service that was designed with them in mind. most experiences you would probably find with the sighted and audio description would be how to turn it off if they activate it by accident somehow. (which is an impressive thing to do) That's mostly what I find when I look up information about accessibility in mainstream products. When my sister turns it on by accident, it's fascinating for a moment, but she thinks it's kind of creepy. (her words, not mine) As far as the scene in Halloween goes, with scenes like that, it would be great to know about all of the details that make it so incredible, but I guess I can handle a music score with no dialog and general descriptions of what's happening. (until someone comes up with a way to portray that in a short amount of time) I wasn't really thinking of it as a replacement either. Just an alternative that would be easier to create. Thanks for the tip about YouDescribe. It sounds familiar, but I'll check it out again.
Audio description has been found to be beneficial to people with autism who might not pick up on visual cues for body language and facial expressions. As I've already said, sighted people are beginning to watch audio description while they do housework or drive, and they think of it like an audio book.
When you consider that the cost of audio description might be over $20,000USD for a full-length movie, you can't deny that smaller studios will struggle to see the point of paying for auio description as the return on investment is likely to be very small. That's why we need to campaign for audio description. Once sighted people see the benefits for them, they will be more willing to understand its usefulness to us.
I wish the first Halloween film were available with AD. It has a fantastic soundtrack. If the AD were able to convey the creepiness of that guy's look, it would still be an enjoyable film.