Talking Pixels from Outer Space

It’s been an eventful week; We announced that New Eden Station would go live in a couple weeks, we began to assemble our administration team, and we did some much needed maintenance on our websites. Most importantly though, we finally were able to implement text-to-speech!

A little bit of history on text-to-speech within New Eden: around two years ago, our then-host JaggerEstep was able to implement text-to-speech on Baystation code. The result? New Eden went from a quiet server barely peaking at 20 players, to a powerhouse consistently seated at the top of the hub. Everyone absolutely loved the silly robotic voices, and while some of the initial hype eventually died down, we were left with a greatly expanded playerbase. Unfortunately, we ran into some unforeseen problems: the program used to generate the speech files began to consistently crash the server. While crashing once in awhile was tolerable, the issue became more and more pronounced, and we were forced to remove our prized feature.

Jump forward to a couple months ago, back in September. JaggerEstep, who had left the SS13 community long ago, came back bearing a gift: a re-engineered voice code wrapper! Surprisingly, nobody seemed to take even slight notice. I quietly downloaded the program, only to discover that it required the correct libraries to use. Despondent, I sat on them for months.

Fast forward to last week. Being struck with inspiration to reboot New Eden Station, I quickly began researching the text-to-speech code that I had received from Jagger. While acquiring the libraries took only a few hours, integrating the program into Space Station 13 took a little more effort; I sat at my laptop, coding in three separate languages, for fourteen hours straight. Four PM to Six AM. I was a man possessed. Thankfully, I inspired my roommate to help me, and he provided some key insights into solving the problems that voice code presented to us. Overall, it was an extremely rewarding experience to get it working, and wonderful practice for future programming endeavors.

And here we are now; the trial of text-to-speech proceeded according to plan, and thanks to some very thorough testers, we discovered some key bugs in the initial design. The testers were also incredibly inspirational and thought provoking: watching them and talking to them gave us some great ideas for how we can further expand this concept.

A few things you can expect to see in regards to text-to-speech:

  • First, let me address the elephant in the room; you will be able to disable text-to-speech if it isn’t your style. We don’t want people to be forced to listen to something they don’t want to hear. OOC voices and deadchat voices will also be turned off, as we found these to be disruptive to gameplay.
  • You’ll be able to change your voice. Before you get super excited, let me break the bad news: there’s only eight or so voices to choose from. However, eight is a lot better than listening to the same one every time. Also, you’ll be able to edit your rate of speaking, so theoretically we could get lots of variation beyond just eight different voices.
  • Alternative language support will be added and utilized for special circumstances. We don’t know where we even might go with this yet, but we’ll find something. You won’t be able to choose different languages for your character.
  • Singing will be supported. Currently, singing requires the user to enter a complex and ugly series of commands and phonemes (word of the day right there). What we’re planning on doing is pre-generating the sound files for a large list of songs, and users will be able to use specific verbs to sing. This feature will have a high potential for abuse, so I expect that we’ll be working a lot towards fine-tuning it. The current system of entering voice commands and phonemes will also be disabled, as they are just way too disruptive to have in IC chat.
  • Other pre-generated sounds will be implemented in conjunction with the current system of preset emotes where applicable, and we’ll be adding some of our own as well. You’ll be able to use these sounds with the click of a button. It’ll take some research to see how many different emotes we can represent, and it may not be that many, but I promise that you’ll see at least a few.
  • Various bug fixes to make text-to-speech a cleaner overall experience.

I hope you’ve enjoyed the first (and rather lengthy) blog post! With the premier of New Eden Station in a couple weeks, we have a lot of preparation to do; I can’t wait to see what we can turn out by launch time. Thanks again to everyone who helped test text-to-speech, as none of this would be possible without you all!