Djinni and Text to Speech Programs

[username_2080208]

Djinni and Text to Speech Programs

I an currently looking at this possibility and was wondering if anyone else had done any research in the area. Some of the Text to Speech programs available today are very good vocial quality and there should be a way to do this.There are numerous such programs available in all release types, IE.. Free Ware, Share Ware, Commercial and Open Source. What I am attempting to do is hook one in with the Djinni editor, with a Plugin if possible, to generate Voices in Mods The source for the conversion will be from the Djinni dialogs, and once the .talk files are created, they can be packaged and distributed with the Mods. This should require only the mod developer to have the Text to Speech Plugin for Djinni. Only problem that I can see on the mod distribution end, it is the size of the .tlk file if a large amount of Text to Speech is done. Syncing the speech to the animation is way beyond where I am at now and, beyond this comment, I am not even thinking about it. :wall:As I have said, I am just starting to look into a way to do this, but I am a 'HACK AT IT' type programmer and there are probably many much better ways to do it then the way I am going.I have not seen anything on this subject posted anywhere on the web. I am hoping that some of you folks might know of links to articles, globs or programs that might work in this direction any of your constructive feedback with ideas I truly appreciate.( This is something that sould work for all modifiable games (.mod) and would add more depth and scope to them. The immersion factor, Hearing rather then reading the Dialog is the main thing that is missing from most user created mods to date. )

 

[ailinon]

Export all dialogue lines to text files.Hook up some sort of automation to create .WAV or .MP3 files from each of the lines.Link the resulting audio files to text lines in Djinni.Done.The quality, however, will be decidedly appalling, compared to anything human, and I'd wager most players will frantically mash the "disable all sound" button within mere moments, due to uncontrolled haemorrhaging from the ears.

 

[username_2080208]

LoL...Actully Ailinon, The Text to Speach programs are much improved over the original Microsoft Sam voices. There are several in the comercial domain that are difficult to difrentuate from human speach when used in simple 2 or 3 word phrases.But, as to your sugestion, I have tired that, and it works to an extent, however the methoid leaves much to be desired. I want to do it all from inside the Mod editor. Do a search for "text to speech synthesis demo" and remember to keep your demo sentance short and simple.take a moment to Listen to the demo at

 

[ailinon]

As a matter of fact, speech synthesis and analysis were parts of my curriculum - so I believe I am quite aware of the issues involved and the current progress of such projects However, your approach still presents two major obstacles:1. You insist on some sort of plugin development, for a program that supports no such notion,2. While speech engines can be quite successful for very short sentences, they fail miserably over longer passages of text - becoming monotonous and, finally, insufferable.Although the former can be, as I have outlined, worked around - the latter I believe to be a solution satisfying perhaps only for sight-impaired players.

 

[username_2080208]

Speech Synthesis and Analysis were also a part of my schooling, Circa 1992. So this proves that while training is nice it by no means indicates that I am current in the field. However, that being said, and the fact that you are the only one to speak with on this subject in this forum tends to force me to answer your statements.First, you put forth " You insist on some sort of plug-in development, for a program that supports no such notion" Sense you can have no idea which Speech to Text program I am working with, you must be referring to the Witcher Djinne engine. This engine is based on the BioWear Aurora engine, and it most definitely does support Plug-Ins to access other moder applications." You insist "?? where did I say that. What I said was " What I am attempting to do is hook one in with the Djinni editor, with a Plug-in if possible " additionally I said " As I have said, I am just starting to look into a way to do this, but I am a 'HACK AT IT' type programmer and there are probably many much better ways to do it then the way I am going. " Now, I did say that "I want to do it from inside the editor" but, that is by no means to say It Must work from inside the editor. Give me a valid suggestion as to how to automate it maybe? Second, you put forth " While speech engines can be quite successful for very short sentences, they fail miserably over longer passages of text - becoming monotonous and, finally, insufferable." Sorry, that is an opinion, not an obstacle. But, if you really want to pick this thing apart, Most game situations, admittedly not all, can be resolved with 5 words or fewer in the sentence. If you run into words that do not sound the way you want, simply use different words. And your last item. " Although the former can be, as I have outlined, worked around - the latter I believe to be a solution satisfying perhaps only for sight-impaired players. " Well, your work around method is a positive input so I will leave it be. But, the last part, about sight-impaired Players, good point, also what about the Reading Impaired players? But then, there are the very small number of game players that simply detest sub-titles (like me) as they ruin immersion in the game itself.Oh, and by the way, I missed this " Hook up some sort of automation to create .WAV or .MP3 files from each of the lines. " by this, would you suggest I use a Text to Speech program that exports to MP3 or Wav formats?But then, I guess I might as well give this up on this forum. I had hoped for intelligent input on how to do something, not a lecture as to why it is not worth doing.

 

[ailinon]

Hey, that's your input right there. If you only expected positive comments, then you should've somehow made it clear in your first post Naturally, from the very beginning I meant that the TtS engine should have an "export to file" function, most expectedly as a command-line option. This way, creating the audio files would be easy as pie, and so would linking them to dialogues in Djinni. I'd wager for a self-respecting TtS engine such an option is a must.Still, while it would indeed be a solution for all kinds of impairments, it would in no way be a definite solution enriching every player's experience - as some players (like me) would immediately kill the audio so as not to hear characters stumble over difficult words. So, while it would indeed be an interesting "emergency" feature, I wouldn't dub it the end of the era of silent mods (pun not intended).