Vega Strike Forums

Turbo wrote:

travists wrote: I have two Cepstral licenses, a male and female, so could easily do text-to-speech computer voices.

I just ran through some demo phrases on http://cepstral.com/demos and it sounds good. It mispronounced some words when I fed it a few VS lines, so phonetics might be required.

I used my own (altered) voice for the translator in ship-to-ship communication* because the text-to-speech that comes with Windows was barely understandable. But the technology is getting there. Even if we leave my work as the translator in place, a Cepstral voice would be great as the voice of your ship informing/warning you, precisely because it should be different from the translator. I like the "Allison" voice because it is easiest to understand, and I always imagined my ship would be female if it had a voice (a Western-culture sailing thing I suppose).
...
Come to think of it, an audio warning for any event I did not initiate, but will affect me, is worth considering. For example:

Urgent events:
"targeted" -- someone locked on to you; maybe hostile, maybe not
"incoming" -- someone shooting at you
"critical damage, eject" (it would also be nice if critical systems like reactor, hull, and shields could give warning when they drop below 50%, can the game do that?)

Not urgent, but helpful:
"SPEC available" -- out of the base gravity well, can SPEC
"SPEC disengaged" -- intered the gravity well, SPEC disabled
"entering jump point" -- press J to jump
"entering docking range" -- press D to dock

I use Diane and David for similar reasons (US English) When I did my own "Spec Engaged" I used the spacetime effect I think it was, for a nice artificial ship voice.
I agree mostly with your list, but might add auto engaged/disengaged. "Incoming" I would think would be a missile launch. As for the eject warning, one of my profs back in college had some fighter-pilot experience about the time they switched from buzzers to voice alerts. For something as critical as an eject warning, he preferred the klaxon to a soft vocal "bailout". Anyway, I have no problem posting voices for the mentioned warnings. Just one question: should I post them all as one audio clip for review, or separate files for easier integration. As for the male-female, that is already handled in the code (config file). Use that as a male/female, or a vocal/buzzer switch?

It makes sense for the eject/bailout warning to be VERY different from the others.

Agree on adding the auto disengagement.

Integration will require one file per event, in OGG format*, so please link to a ZIP with individual files. Thanks!

Edit: Once the samples are done, we need comments from our international audience. People who want to play VS, but may not be entirely comfortable with English, need to be able to understand the voices.

* http://vegastrike.sourceforge.net/wiki/ ... quirements

Turbo wrote:Integration will require one file per event, in OGG format. Thanks!

This is new... the present cocpit sounds are .wav, and I did just fine using .wav (at least a few updates ago they where)

don't have a program to convert to OGG right now, but I'll get the waves done ASAP.

Turbo wrote:Edit: Once the samples are done, we need comments from our international audience. People who want to play VS, but may not be entirely comfortable with English, need to be able to understand the voices.

Perhaps a config setting for international audio? There must be some setting to address this, as there is a lot of text that would need to be translated too. I'll take a look at the link you added too.

Here is what I have so far.

Thanks very much! Actually, submitting them as WAV worked well -- I can highlight the unZIPped files, "Add to WMP list", and it plays them all.

First, what I liked:
The "dizzy-droid" sample was too short for me to hear what the effect was trying to do, so I'd like to hear more of it. I liked "liquid love" because it sounds more subtly mechanical than the others.

Second, what I didn't like (at least, not for use as the ship's voice):

The "PVC pipe" and "old robot" effects sounds too much like the Mechanists we already have. That could be a shortcut to getting a female Mechanist voice set.

The split-personality effect is an over-the-top version of what I was trying to do with the Klk'k in the other thread. If it is possible to dial it down (we don't want demon-possessed, just one beastie with 2 mouths speaking our language), we might have something there too.

Of course, if the program can't emote appropriately to the lines being spoken, then we should not use it for organic voices. Even Mechanists have human brains and emotions.

But, the program sounds (to me, need to hear from the international audience on this) clear enough to be our new translator voice -- we still haven't started working on how the Dgn, Shmrm, and Uln might sound, and will need the translator lines for them. It could even replace my past work as the translator voice for the Aera and Rlaan. I have those alien voices without the translator uploaded to /masters/, so splicing in new translator lines is possible.

The Cepstral voices seem to be flat in regards to interpreting punctuation. It does support several SAPI5 tags though, so one can change pitch, volume, and emphasis. A license is not needed to play with the voices, they just have a sales pitch they go through first. It will take some tweaking to get them to emote, but there are tools to make it happen. Looks like there is the possibility to create a custom effect; as such making the split-personality effect not quite so split might happen. If you don't have hard bandwidth limits download a voice or two and see what you can come up with. Two heads are better than one!

Here is the longer dizzy droid.

Thanks for that. I think either dizzy or liquid would be great for the ship's voice...dizzy being a bit more mechanical than liquid, but neither one too mechanical.

The ship's voice has a limited number of things to say, so high quality would be cheap. The Mechanist pilot is more likely to go cheap on his voicebox and save his money for quality sensors and appendanges that help him fly. So please don't think I'm saying we should change the Mechanist voices we already have.

I encourage you to go forth and create the ship's voice set with one of those voice effects, in male and female versions. I have rights to put stuff in SVN, but unfortunately the game won't play for me now, so I can't help with the testing we need to do before committing to SVN.

As for your suggestion about multiple lines of effort, I agree. Depending how far the ongoing discussion gets regarding the Klk'k, I can start scripting the translator overlays for our 3 remaining aliens until we can decide how those aliens sound.

I was going over the licencing info again at Cepstral.com, and it looks like I won't be able to help after all. I can see needing aditional licences for embeding the voice in a web page, but man... spare an extra $200 us so it's legit?

What exactly do I get when I purchase an Audio Distribution License?
An Audio Distribution License (ADL) provides rights to publish audio created using a Cepstral voice. This includes the rights to distribute audio files containing Cepstral voice output via telephone, over the air and television broadcast, audio CD, or any other physical media. ADLs do not license the electronic distribution of audio via the Internet. ADLs are sold on a per-voice basis. That is, a single ADL gives you permission to distribute the output of a single voice. To distribute the audio of several voices, you must purchase several ADLs - one for each voice.

When you purchase Audio Distribution Licenses on the Cepstral Store (at $199.95 per voice), you will receive an email with instructions on how to access a printable PDF copy of the license agreement. After downloading this license, you must fill in the Licensee fields, list the voices you have purchased the ADL to cover, sign and date the license agreement, then send it back to us, either via facsimile or postal mail. The fax number and address will be provided to you after your purchase is complete.

Still more than happy to help if there is a better source for voices out there.

Ah well, better to find out now, then run afowl later.

**edit: maybe someone better at legal mumbo jumbo than me can contact them. this is a free project after all... **

Wow, I'm glad you caught that now.

Better source for voices? I was playing with eSpeak, but the resultsaren't great.

How about the voice source in your own head? Long ago I wrote a tutorial which covers the basics of "how". If you don't like your own voice for the ship's computer, the wiki has a list of factions not yet voiced.

Funny how this thread started about sound effects, then we sort of talked ourselves into voices instead. My fault, I'm sure.

Yeah this one wanders a lot. did some more digging:

https://ondemand.neospeech.com/terms.php

https://ondemand.neospeech.com/pricing.php

I don't have the funds right now, but It's an option.

Still think that non-comercial use like this should not be an issue, but unless a waver can be done.... Now if someone had the scratch to buy a major distribution licence.... well built in text to speech would be amazing!

Turbo wrote:Better source for voices? I was playing with eSpeak, but the resultsaren't great.

As we are speaking of speech synthesis, Festival may be a better option. It may even be possible to embed it into the game, so that we can have any written text read out loud. However, I have not investigated this in detail.

Just to put a scope on it, I picked 3 factions whose scripts are already written and ran a word count: the militia faction is ~600 words, pirates ~300, and Klk'k ~550.