Why do we have text messages in game instead of voice?

[ozlizard]

Yeah sorry if I missed the part about them having a magic wand they could magically wave to add voice without any additional dev time or capital then my mistake. Im all for the magic wand approach if that is available. Even a text to voice program would need licensing and a lot of coding to implement properly. Just not a priority with all the other things that they could be doing.

What a load of crap. This would be quite a simple thing to implement. Just shows laziness on their part.

But I forget, the game is perfect, how silly of me.....

- - - - - Additional Content Posted / Auto Merge - - - - -

except text to speech typically has terrible cadence and completely lacks emotion and context.

Well, given that the NCPs are not human players, are never even pretending to be human players why can't they be remote computer controlled drones who use computer voices?

Sure beats trying to pretend that they are real.

 

[Blast Hardcheese]

Might be alright to have it in addition to the text, but I find the text to be much more useful than voice would. I am too concentrated in things happening on screen to pay attention to someone randomly talking to me. And like other people said, text-to-speech is really terrible and not a solution in any context.

 

[Valethar]

Because typing is relatively free, and looks the same no matter who does it. If someone quits, moves to a different task, etc.. the user won't notice a difference.

Voice acting is expensive, and you're dependent on their schedule, which means you may not be able to get samples quickly enough when you change or add something to the game. Using someone else tends to be a problem unless they sound a lot alike and/or you can alter the voice with software/mixers.

 

[The_RayGun]

I agree with the OP. And I understand that it would take a lot of time and money to have actual voice actors, but it seems that they could potentially let the ED community contribute a significant portion of the voice content... Just an idea

I would happily donate my voice for this. To do this I would require a list of all the phrases that need to be said, recommendation on recording software (freeware preferably), guidelines on how to save the records (file types, recording settings for software etc) and I would go through and do that.

Times that by enough enthusiastic members from the community and we have a wide range of voices that can be used.

Put the voices through a filter to get them sounding right (which again I could do as saving the recordings, if given instructions on how to do) and if we are lucky we have something that can be used in game to add to the immersion.

 

[Armour]

I made a comprehensive suggestion about this topic a couple of months ago, I even got in contact with some folks in the industry to see how practical it could be, but it turns out that they weren't really on my wavelength.

My idea was to record the speech markup (timing/emphasis/pitch) information from someone reading the text pieces, but slot in different timbre data at runtime, to vary the actual playback voice. If that timbre data is understood well enough, it could even be procedurally generated by mixing multiple data sets, so that each voice is unique. So they would be recording the essential parts of the voice data, but throwing away the original speaker's voice. The storage savings would be immense, and the quality would be a lot better than standard text-to-speech.

This would allow them to burn through the phrase database quickly with a bunch of cheap voice artists. No worries about who is off sick or working somewhere else at what time, no rights issues to content with.

And with some of the techniques used in modern IF parsers, one could probably slot together the sub-phrases in a natural sounding way so that new sentence combinations could be generated on the fly. The only real issue I can envisage would be performance constraints, but I would think that a lot of the stuff that a real IF parser has to contend with could be stripped away so that the playback could be optimised somewhat. No such worries about the recording side, that can be as CPU-intensive as necessary to ensure accurate results.

This technique could be used for all voice interaction with NPCs, for news bulletins, and would be flexible enough to be useful in a variety of other situations (and even other games).

But I agree that it's not a priority at present.

 

[Darkwolf]

My idea was to record the speech markup (timing/emphasis/pitch) information from someone reading the text pieces, but slot in different timbre data at runtime, to vary the actual playback voice. If that timbre data is understood well enough, it could even be procedurally generated by mixing multiple data sets, so that each voice is unique. So they would be recording the essential parts of the voice data, but throwing away the original speaker's voice. The storage savings would be immense, and the quality would be a lot better than standard text-to-speech.

That's a really good idea. Because the phrases are fixed, a much better job can be done of handling that kind of data than what a generic text-to-speech converter can do.

Course, it does mean that all the voices will come across fairly similar in the end. But at least you won't get the jarring effects you get right now with how badly text-to-speech handles regular text.

 

[Armour]

That's a really good idea. Because the phrases are fixed, a much better job can be done of handling that kind of data than what a generic text-to-speech converter can do.

Course, it does mean that all the voices will come across fairly similar in the end. But at least you won't get the jarring effects you get right now with how badly text-to-speech handles regular text.

I omitted one thing here that was included in the previous incarnation of this proposal - which was to vary the playback prosody (the pitch/rhythm/emphasis elements) with a certain amount of a procedural signature for each speaker. That way, some NPCs would speak a bit slower, some with a bit more extreme expression, some with a bit more of a sing-songy pitch envelope. It would be slightly more difficult to apply the criticism that everyone could come across fairly similar if that were also implemented, I think.

 

[TheNut]

I don't think you don't need voice actors for a game like Elite. There's no drama to convey, no protagonist or antagonist with lengthy dialogue. Just simple radio chatter or banter. You already hear some of that ambience when you dock at stations and there's a few voice overs from the ship's computer. As long as the person is somewhat capable of immersing himself into the role, once you add in the radio filters and other special effects, it can produce a good result. I'm pretty sure there's a few capable people at FD that could help contribute to adding voices and it would only take a scheduled lunch or two to sit with their audio engineers and have some fun recording voices for the game.

I'm also not against using a text-to-speech engine for computer dialogue that would be difficult to support real-time voice overs for such things like galnet news or the marketplace. Since I like to multi task, I'd rather the computer read out the information to me while I'm busy trading goods or flying to a new sector. With a little filtering and reverb in there, it would be passable for me. There's a few really good text-to-speech engines out there too.

So yeah, more immersion = good in my books! Bring it on!