Open public beta for the NPC speech synthesis add-on 'TTS4ED'

Cosmo

Cosmo
Repost by demux4555.

Hi there,

I've had the chance to get rid of the most critical issues and bugs, and I now feel it's robust enough for public use.
What's this TTS4ED all about? Example video here, and previous reddit discussion here.

The official page is here: sites.google.com/site/tts4ed (downloads, installation, etc)
The official support/feedback subreddit is /r/TTS4ED/

I hope you enjoy it as much as me.

If you are updating from a previous version, make sure you delete the old exe (it has a new name now). It is safe to overwrite settings.ini (recommended), as long as you remember to enter your accountID/password again. Do not delete the download folder, unless you are prepared to use more CereProc credits on downloading the speech again.

Note: I just want to make it clear I am in no way affiliated with CereProc, even though I have made a discount deal for purchased credits.

[video=youtube;syRE_nClBS0]https://www.youtube.com/watch?v=syRE_nClBS0[/video]
 
Last edited:

Jack Schitt

Jack Schitt
I knew someone would have a crack at this some day. :D

Jack Schitt said:
Because NPC communications are now recorded to the Journal e.g...

Code: 
{ "timestamp":"2017-01-16T19:28:23Z", "event":"ReceiveText", "From":"$npc_name_decorate:#name=Chris Armstrong;",
"From_Localised":"Chris Armstrong", "Message":"$BountyHunter_HunterHostileSC_Relevant03;",
"Message_Localised":"There you are! That bounty makes you today's meal ticket.", "Channel":"npc" }


...it should be possible to write a program to parse the file in real time, extract the actual dialogue and send it to a text-to-speech engine. If you wanted to be clever you could compare each NPC name with a database of gender-weighted names and allocate a male or female voice accordingly. If you wanted to be really smart you could use a hash of the NPC's name as a seed for picking from a selection of voices, thus ensuring that the "same" NPC had the same voice each time you encountered them.

This is well beyond my meagre string-slicing skills but it seems like an ideal project for an enthusiastic coder to give it a shot if FD don't want to. Hell, if someone's already done it please let me know; I would love such a feature and a donation from my PayPal account stands ready... ;)
Click to expand...
Although I didn't expect it to be quite so soon or so thorough to be honest. It looks really good. It's a pity the choice of TTS engine is likely to incur a nominal cost, because I'm reasonably certain Amazon's offering would have absorbed most average ED players' traffic into its free tier. But maybe the range of voices wasn't as good.

Looking forward to having a play with the program this evening.
 

rucky

rucky
very interesting! I will check that out!!!

but I wonder if it's not possible to make some compilation of the most used pirate/bounty hunter phrases on our own.
25 pounds isn't in-expensive and I'm not sure how many such packages are needed to get (mostly) all phrases done...
(perhaps not possible because redistributing is against copyright law?)
 
Last edited:

Comrade Napoleon

Comrade Napoleon
rucky said:
very interesting! I will check that out!!!

but I wonder if it's not possible to make some compilation of the most used pirate/bounty hunter phrases on our own.
25 pounds isn't un-expensive and I'm not sure how many such packages are needed to get (mostly) all phrases done...
(perhaps not possible because redistributing is against copyright law?)
Click to expand...

As cool as it sounds the 25 quid thing will probably cause an massive otrage somewhere...
Remember the name plate tantrums for 1-2 quid for 9?
The sub thing is not me, really, but still good idea, and yes development and all that costs, and Dragon speech is not free either.
Anyway I appreciate all stuff that CMDR's do even if some cost some .

Cheers Cmdr's

Edit: and Thx CMDR Cosmo for the heads up.
 
Last edited:

rucky

rucky
Comrade Napoleon said:
As cool as it sounds the 25 quid thing will probably cause an massive otrage somewhere...
Remember the name plate tantrums for 1-2 quid for 9?
The sub thing is not me, really, but still good idea, and yes development and all that costs, and Dragon speech is not free either.
Anyway I appreciate all stuff that CMDR's do even if some cost some .

Cheers Cmdr's

Edit: and Thx CMDR Cosmo for the heads up.
Click to expand...

yes, that's why I asked for making a community phrases compilation ;-)
30€ (and then it even does'nt suffice) is an amount I can't spend on such gimmicks... sadly... :-(
(but could be against law, I'm not sure)
 

xondk

X
NeilF said:
Why aren't FD all over this with their own version/implementation?
Click to expand...

Because as much as people like it quality is not really up to production standards, plus TTS is generally third party, and developing a proper TTS system/engine, is something they could do, it would be difficult to bring it up to a certain quality standard.
Since this is third party they do not need to stand for the quality, I think TTS is fine, and the concept is a good idea, without a doubt, but over longer time, I'd get annoyed at it, lack of flexibility and such, at least in this current concept.
But if FDev wanted to work on it and try to develop something proper, that could be cool.
 
Last edited:

Toasty

Toasty
fantastic effort. I'd love to see something like this professionally implemented though. certainly without the requirement for 3rd party accounts. excellent work though.
 

atlas974

atlas974
I'm using it and it improve a LOT the immersion while playing in VR (no need to try to read in middle of combat what npc are saying). Voice variation is great (30+ differents voices i think) and once they are processed by the cloud server they are store on your computer, so what is processed won't be taken in the 10000 free characters offer when you sign up (and you got 10000 each month)
 

iFred

I
First create a CereProc website account and then visit the CereVoice Cloud registration page. A free 10k credits developer account will get you started so you can get an impression of the TTS, but after spending the initial free 10000 characters (should take a few hours of gameplay, I'd guess), you need to buy buy credits unless you want to wait another month for a credit refill. I have made a deal with CereProc for a 10% discount using the coupon code found in the readme.txt included in the program download. Note that the synthesized speech is cached, so I'm quite sure 100k credits will be sufficient, unless you use the program for TTS on incoming friends/player chat as well. Why pay for TTS? Because if you want convincing and quality speech synthesis with lots of different voices, you'll have to pay for it. It's that simple. Also, ~20GBP is not very expensive, in my opinion.
Click to expand...

Good idea - but I think I would pay 20 GBP to FD instead of a TTS-Thingie.
 

knowles2

K
NeilF said:
Why aren't FD all over this with their own version/implementation?
Click to expand...

Because they don't think the technology is mature enough to sell to the masses.

It not like David Braben hasn't kept track of developments in this area, he was a investor in Phonetic Arts which was acquired by Google an at the time in 2010 was consider one of the best speech synthesizers around.

Wavenet from Deepmind shows that procedurally generated voices that sound human aren't all that far away, another few years, through someone like Frontier taking on this challenge could speed that up.

https://deepmind.com/blog/wavenet-generative-model-raw-audio/
 
Last edited:

sonichi813

S
Found this yesterday due to Obsidian Ant's video. Awesome tool and with 10,000 cr per month, it might actually be free of charge. I played for about 2 hours today and only spent 300 cr. It really adds to the immersion and atmosphere of the game, volume can be adjusted and response time gets better as you use the tool. I for one can't wait until something like this is implemented into the game, but until that happens I'll be using this. Thanks & fly safe 07.
 

Andovar

Andovar
The stuff that the community comes up with never ceases to impress. [up]

Here's to hoping this is on the Audio team's to do list. :)
 

bookmite

bookmite
I watched the video, looks amazing and thank you for your efforts in helping the community.

A lot of cmdrs that use VA also have the viona voice packs, I have 3 of them, is it not possible to utilise your own voice packs ?

I'm hoping that this idea is implemented in game as it will add great immersion. Or perhaps VA wil add it as a feature of their own program ??

thanks again, downloading now
 

Cervantes01

Cervantes01
FD should be ashamed that this type of thing isn't in the game already, it's a glaring omission if you ask me. Well done on this, superb tool.
 

evilscoop

evilscoop
wow...
As a user of hcs voicepacks, and also a user of VR, this is one of those tools that adds to the immersion....
Yes id prefer voicepacks than the 3rd party tool...but hey, its a beta...its early days....
 
You must log in or register to reply here.
Back
Top Bottom