Audio Looking for a free, compatible Text-To-Speech system

I've been working on a project and for now simply need a free TTS system that can be used as a prototype for Unity 2018.3.6 on Windows. The problem is that I have found it very hard to find any current and compatible systems that work with Unity.

After looking on Google for quite a long time, I found several dated posts, but none of which work or are not crafted completely from scratch (I understand that for the end goal this will be needed, but for now time is short and we simply need something good enough for a prototype).

I've tried using System.Speech, but that has reference and compatibility issues. ISPVoice also has issues. I've found someone's work-around for this one, but the problem is the work-around has very minimal functionality. It does not allow voice modification, which is sort of a necessity in order to differentiate speakers, nor much else. It simply speaks, with a possible delay.

Does anyone know of an already-built system? A basic one is totally fine, as long as voices can be modified somewhat.

 

Seconded. I'm studying a masters in Serious Games, and I want to produce a language game for my dissertation project. I've got access to several SAPI voices free, but that ain't gonna be much use if I can't get them into Unity.

 

Apparently there is an Android TTS that can be used in Unity, but I believe that is more on the Android side rather than the Unity side. I suppose I can understand why it wouldn't be a commonly used thing, since the voices for TTS are not really that good audio wise, but it would still be a nice feature.

 

Well, given that there are laws in many countries saying that computer software should not discriminate unnecessarily against people with disabilities, and given that OS-level TTS is basically now a freebie that's got massive benefits for people with visual problems, the various engine makers would do well for their customers (i.e. game/app developers) if they built in support for it at ground level....

 

Well, Game Engines are not really designed for the blind, so TTS would not fall under such discrimination laws. It would not be possible for a blind person to develop a game through a Game Engine, at least not without the Game Engine having a specific mode implemented that would audibly describe EVERY action being performed and...well, much much more. You would basically be creating an entirely new game engine focused specifically towards that aspect.

Suffice to say, TTS being added to Unity would be a nice addition, but is in no way something in high demand right now. High-Quality games will use pre-recorded audio until TTS systems with excellent quality and high-modular aspects are developed.

For now, the team I am with has decided to implement Amazon Polly. We had to do some searching for a github to use it, and then had to modify some things, but it works well enough for now.

 

I don't think that's how discrimination laws work:
Judge: "Your software doesn't work for the blind, thus you are charged with breach of the disability discrimination act."
Defendant: "My software isn't designed for the blind."
Judge: "Oh, I see, case dismissed."

I'm being frivolous of course, but it remains true that there are people whose vision or visual processing (or even just dyslexia) makes reading difficult or impossible, but still have enough vision to play games. (Perhaps not the latest high-speed combat games, but slower adventures etc.)

For an indie developer who can't afford professional voice actors, text-to-speech is an adequate way to at least offer something to the visually impaired, who will already have high-quality TTS software installed on their PCs and phones anyway. Unity is used by a lot of indies.

Cool. Sadly for me I'm trying to make a game in a language Amazon don't support, but for which there is a SAPI voice available. :-(

 

I am no expert in this field but I had tried to create one using fastspeech TFLite model. I wanted to use Unity's barracuda inference engine but it sadly did not support transformers yet...
any help or feedback would be really appreciated!
https://github.com/voxell-tech/UnityTTS