ROM Hack Tomodachi Text to speech analysis/mod?

[XDM]

so now that im digging around in 3ds game files i was always wondering about replacing the Text to speech engine in tomodachi life.

upon digging around in the files i found the folder thats responsible for the engine.

a folder called libNTTS.
so far
inside the components_cmn_LZ.bin
are 2 files
lid.dat and synth_med_fxd_bet2f16.dat

from what i gather is that lid.dat is all the voice data and the synth_med_fxd_bet2f16.dat is the module that alternates pitch, speed and variation of the voice.

the components_enu_LZ.bin
holds
clc_enu_cfg1.dat
clc_enu_samantha_cfg1.dat and
select_samantha_bet2f16.dat

searching that up shows that tomodachi is most likely using a known TTS engine called "depes tts"
link Link2

the dictionary_enu_LZ.bin holds a userdct_enu.dat that when opened in notepad++ are some words hence the name dictonary_enu
the header_LZ.bin is not a container file but is instead directly read, it seems to be the header code/config for the engine as well as something to do with "British English-Serena (TTS)"

and finally the last file is ruleset_enu_LZ.bin
which holds a userruleset_enu.rst
that simply has this in it

[header]
language = ENU
charset = utf-8

[data]
/\b(Who|Whom|Whose)(\s+[^.!]*?)\?/   -->   $1$2.
/wii\-u/i   -->   "\\mw\\Wii U\\mw\\"
/fils aime/i   -->   fils-aime

 

[duwen]

Interesting... any chance of replacing the text to speech engine with an NTR plugin?

 

[XDM]

Interesting... any chance of replacing the text to speech engine with an NTR plugin?

im not sure, i haven't used NTR for much.

 

[duwen]

Something I've wondered about with regards the speech in Tomodachi Life; the English speech in the EUR version has a definite British accent - is it the same in the US version, or does it have an American accent? If it's the latter it could be possible to 'patch in' other regions accents.
Similarly, when you have an islander from somewhere like Japan or Korea visit your island it makes no attempt to translate names or messages via the text-to-speech - would it be possible to combine foreign dictionaries with the stock version to fix this?

 

[Dr.Hacknik]

It would be cool to replace present One's with a more Realistic Voice.....It would probably be a little hard to port over though.

 

[XDM]

like Japan or Korea visit your island it makes no attempt to translate names or messages via the text-to-speech - would it be possible to combine foreign dictionaries with the stock version to fix this?

Don't think so seeing how there only one set of language files for each ver. It's one or the other as far as I know

 

[XDM]

I may be pushing it but I want to replace with the windows tts. Or maybe the android os ver seeing how their simular.

 

[LeafyIsHere]

Anyone find out any more about this?

 

[Calafska]

Anyone find out any more about this?

edit : It would be so nice for our rom-hack

damn im to tired

 

[XDM]

I would love to replace the engine still. The android tts is pretty good but I'll have to decompile the android engine as well

 

[LeafyIsHere]

I would love to replace the engine still. The android tts is pretty good but I'll have to decompile the android engine as well

Yeah, I think the Google tts voices are higher quality than what comes with the depes engine.

 

[Calafska]

Yeah, I think the Google tts voices are higher quality than what comes with the depes engine.

yes , I cant argue with that
I would love to do that ^u^

 

[LeafyIsHere]

@xdarkmario @Calafska
Cool, I just figured out what exactly those .bin "ENU" files are, and how they can be swapped out for higher quality versions.

Tomodachi Life voices are created using low-quality versions of TTS voices from Nuance.
I found a site that allows you to download a demo of any Nuance TTS voice (all are high-quality).
Surprisingly, the computer TTS files are the same file-format as the Tomodachi Life speech files. (except they are HUGE compared)
Here is my Tom US voice:

I think someone COULD replace the 3DS Tomodachi speech files with the high quality ones,
but the game would take up like 2 gigabytes.

 

[XDM]

@xdarkmario @Calafska
Cool, I just figured out what exactly those .bin "ENU" files are, and how they can be swapped out for higher quality versions.

Tomodachi Life voices are created using low-quality versions of TTS voices from Nuance.
I found a site that allows you to download a demo of any Nuance TTS voice (all are high-quality).
Surprisingly, the computer TTS files are the same file-format as the Tomodachi Life speech files. (except they are HUGE compared)
Here is my Tom US voice:
View attachment 75825
I think someone COULD replace the 3DS Tomodachi speech files with the high quality ones,
but the game would take up like 2 gigabytes.

Interesting, Good find. In this age of mass storage 2gb are nothing. I might try this

 

[LeafyIsHere]

Interesting, Good find. In this age of mass storage 2gb are nothing. I might try this

btw if you do try, Serena UK is the voice for the EUR version, and Samantha US is the US version's lol.

 

[XDM]

also theres a file in there called
clc_enu_samantha_cfg1.dat

goggling that i found a source code

Github

and that points to some old android Motorola tts files.
which is kinda odd because thats another engine.
as well as another company called
Nuance, a (TTS company)

see here for some of there work. if we can at least replace it with this this would be cool
http://codefactoryglobal.com/speech-technology/voice-licensing/
dosent look free but its a example

 

[LeafyIsHere]

also theres a file in there called
clc_enu_samantha_cfg1.dat

goggling that i found a source code

Github

and that points to some old android Motorola tts files.
which is kinda odd because thats another engine.
as well as another company called
Nuance, a (TTS company)

see here for some of there work. if we can at least replace it with this this would be cool
http://codefactoryglobal.com/speech-technology/voice-licensing/
dosent look free but its a example

Hmm, I am almost certain almost any SAPI TTS voice would work,
I have been researching TTS voices and how they work, and it seems like almost every voice is the same file-format.

All that we need now is a way to unpack and repack the .dat files in libNTTS.
The EveryFileExplorer only unpacks them, sadly...

 

[XDM]

Hmm, I am almost certain almost any SAPI TTS voice would work,
I have been researching TTS voices and how they work, and it seems like almost every voice is the same file-format.

All that we need now is a way to unpack and repack the .dat files in libNTTS.
The EveryFileExplorer only unpacks them, sadly...

well if we cant unpack them then maybe we need to inject mods into the files.

 

[LeafyIsHere]

well if we cant unpack them then maybe we need to inject mods into the files.

Not sure how that would be done.

 

[XDM]

Not sure how that would be done.

I assume we would have to know the memory maps. Similar to how early smash bros mods were I think