How language translators could work

[Spacescifi]

 

I once saw a scifi short that seems a bit more realistic than the translators from star trek.

It is here below.

 

https://m.youtube.com/watch?v=mHLYuOFPf_c

 

The one limit I can think of is voice detection. In a room of people speaking Japanese how do does the device know who to listen to if you only want to speak to one person and you speak english?

What would be your solutions? Since devices are not good at reasoning things out.

My solutions: Free speech rings: Just walk into a ring spot on the ground where your device will only hear the person who joins you. Many other things could be this way, yables, chairs, ect.

 

What are your solutions?

Edited August 9, 2019 by Spacescifi

[kerbiloid]

It should:
Detect all speech sources in the room.
Translate every of them if the language is recognized.
Constantly measure the eyes direction and blood pressure below the ears to define the current attention focus.
Trace the attention focus movement to avoid its random fluctuations.
Define the current object of interest.
Amplify the OoI translation compared to other voices.
Amplify the OoI original speech louder than others but quieter than its translation.
Add a half-second delay between the original voice and the translation, to let the listener listen the speaker's intonations anb tembre.
If HUD is available, show subtitles if the option is on.

Edited August 9, 2019 by kerbiloid

[magnemoe]

1 hour ago, kerbiloid said:

It should:
Detect all speech sources in the room.
Translate every of them if the language is recognized.
Constantly measure the eyes direction and blood pressure below the ears to define the current attention focus.
Trace the attention focus movement to avoid its random fluctuations.
Define the current object of interest.
Amplify the OoI translation compared to other voices.
Amplify the OoI original speech louder than others but quieter than its translation.
Add a half-second delay between the original voice and the translation, to let the listener listen the speaker's intonations anb tembre.
If HUD is available, show subtitles if the option is on.

A bit over complex. An language translator would be limited by default, yes having an video feed would be very nice however its lots of settings it would not work. 
It must be told its setting or be able to recognize it automatically like grouped around an table in an meeting or diner or cafe. 
People standing around here some will speed directly to you or to an group facing each other. 
Recognize others who are important, here language recognition will solve most of it. Is the guy talking loud giving relevant information or just drunk and arguing :), if quite translate anyway. 

And an translator will would never be perfect, it would be good enough for most technical stuff even social. Not so much if you need to convince the other party like an sale as you will miss cultural stuff. 

[Spacescifi]

4 hours ago, kerbiloid said:

It should:
Detect all speech sources in the room.
Translate every of them if the language is recognized.
Constantly measure the eyes direction and blood pressure below the ears to define the current attention focus.
Trace the attention focus movement to avoid its random fluctuations.
Define the current object of interest.
Amplify the OoI translation compared to other voices.
Amplify the OoI original speech louder than others but quieter than its translation.
Add a half-second delay between the original voice and the translation, to let the listener listen the speaker's intonations anb tembre.
If HUD is available, show subtitles if the option is on.

 

Quite complex, and all the characters would have to wear google glass cameras or something. Still interesting though.

That said, what you made is like this:

Kerboloids Ool language translator glasses: When you absolutely, positively, have to talk to everybody in the galaxy!

Also good for for professionals working on a starship.

2 hours ago, magnemoe said:

A bit over complex. An language translator would be limited by default, yes having an video feed would be very nice however its lots of settings it would not work. 
It must be told its setting or be able to recognize it automatically like grouped around an table in an meeting or diner or cafe. 
People standing around here some will speed directly to you or to an group facing each other. 
Recognize others who are important, here language recognition will solve most of it. Is the guy talking loud giving relevant information or just drunk and arguing :), if quite translate anyway. 

And an translator will would never be perfect, it would be good enough for most technical stuff even social. Not so much if you need to convince the other party like an sale as you will miss cultural stuff. 

 

My main scifi use was for aliens/humans on a starship together who do not speak the same language. If your starship medical doctor is a lizard man with a translator, knowing exactly what he is saying becomes vital. As are such things as "Raise shields!" and etc.

Edited August 9, 2019 by Spacescifi

[kerbiloid]

I didn't say glasses are required anywhere. I said "if".
But how should one listen the translation without wearing some device, not necessary glasses?

[Spacescifi]

8 minutes ago, kerbiloid said:

I didn't say glasses are required anywhere. I said "if".
But how should one listen the translation without wearing some device, not necessary glasses?

 

Cameras mainly to track eye movement and object of interest.

All things considered, I think a floor location based translator would be the simplest to implement and therefore the most common.

Stand in that circle there to talk to that other person over there. Wanna talk to random people? Wave at them and have a network of lines with rings painted on the floor to stand within. Use a phone device to specify which ring is occupied by the person you wanba talk to and communicate. Simple.

Also for the big important guy in charge, his voice is like a walkie talkie broadcast that everybody hears in their own language regardless.

Edited August 9, 2019 by Spacescifi

[kerbiloid]

Cameras may be mounted in walls and trace everybodies eyes remotely.

[Spacescifi]

5 minutes ago, kerbiloid said:

Cameras may be mounted in walls and trace everybodies eyes remotely.

 

True, but that is so not happening on an away mission on some alien world talking to each other outside.

[magnemoe]

23 minutes ago, Spacescifi said:

 

Quite complex, and all the characters would have to wear google glass cameras or something. Still interesting though.

That said, what you made is like this:

Kerboloids Ool language translator glasses: When you absolutely, positively, have to talk to everybody in the galaxy!

Also good for for professionals working on a starship.

 

My main scifi use was for aliens/humans on a starship together who do not speak the same language. If your starship medical doctor is a lizard man with a translator, knowing exactly what he is saying becomes vital. As are such things as "Raise shields!" and etc.

Glasses is a bit of an overkill, earbuds should work, stereo give them location of target, glasses with camera could also get body language and better localisation. 
Glasses could also use argumented reality to translate text and symbols like the watch your tail symbol on automated doors. 
Have translator also try to get tone, incoming, raise shield now and brace for impact, or we should raise shield soon we pass trough the ring in an hour. 
Technical communication is pretty easy as it uses an standardized language even if language is different. 

Just be aware of differences, aliens used hydraulic and pneumatic computers for 100 year before they invented electronic. 0 is on and 1 is off  
 

[Spacescifi]

5 minutes ago, magnemoe said:

Glasses is a bit of an overkill, earbuds should work, stereo give them location of target, glasses with camera could also get body language and better localisation. 
Glasses could also use argumented reality to translate text and symbols like the watch your tail symbol on automated doors. 
Have translator also try to get tone, incoming, raise shield now and brace for impact, or we should raise shield soon we pass trough the ring in an hour. 
Technical communication is pretty easy as it uses an standardized language even if language is different. 

Just be aware of differences, aliens used hydraulic and pneumatic computers for 100 year before they invented electronic. 0 is on and 1 is off  
 

 

Yeah, if an electromagnetic pulse fried the away team's communicators that would be a fun episode, something star trek rarely ever explores, and doubt they ever did between crew members themselves.

It would be amusing to see crew flail around helplessly speaking gibberish until they finally start drawing symbols in the dirt with stcks.

Low tech and kinda anti-trek. Funny. Ironic too, and I love irony.

Edited August 9, 2019 by Spacescifi

[magnemoe]

10 minutes ago, Spacescifi said:

 

Yeah, if an electromagnetic pulse fried the away team's communicators that would be a fun episode, something star trek rarely ever explores, and doubt they ever did between crew members themselves.

It would be amusing to see crew flail around helplessly speaking gibberish until they finally start drawing symbols in the dirt with stcks.

Low tech and kinda anti-trek. Funny. Ironic too, and I love irony.

You might need translators simply because the two species was unable to pronounce each other language, could easy have some specie who could not speak a lot of sounds but could echo locate so they language was pretty much modulated Morse at 100 Kz. 

[Vanamonde]

Thread moved to the Lounge, since it's not so much science-ish. 

[kerbiloid]

2 hours ago, magnemoe said:

Glasses is a bit of an overkill, earbuds should work, stereo give them location of target, glasses with camera could also get body language and better localisation. 
Glasses could also use argumented reality to translate text and symbols like the watch your tail symbol on automated doors. 

In future they can just remotely stimulate eye and ear nerves.
So if the living space is full of cameras connected into an optical network (watching walls), no need in any kind of devices in/on the body at all.

And I guess this will happen sooner than a universal translator.

Edited August 9, 2019 by kerbiloid

[YNM]

I personally still think that there are going to be subtle nuances which wouldn't work in instant translation. Languages evolves in each of their own ways, and we'll often miss things because it's just not obvious or it's only recent.

 

Then again automated/machine translators are a helping tool since pronounciations can be incredibly difficult. The problem shifts to "how well can they emulate them".

[purpleivan]

Why not just translate all speach detected, but play the translated audio at the same volume as the orginals being spoken (e.g. someone on the other side of the room would sound quiet, unless they started shouting). That way it would sound like you're in a room of people speaking your language, and people deal with that fine.

Probably need to apply some kind of accent to each speaker though, so it doesn't sound like a room of full of clones... that does complicate things though, as that would require identification of different speakers.

Edited August 11, 2019 by purpleivan