Starting a code project to make my doll speak

timetraveler1 · Post by **timetraveler1** » Wed May 09, 2018 9:04 pm

what you are talking about with voice and custom talk etc. can all be done with the ultra hal a.i. program (pc) i use it and have a wireless speaker near my angela doll , you can create brain plug ins as i did for words that trigger custom conversation . i have done that with x rated talk . write anything that i want my angela doll to say . plus being a a.i. program she actually learns from things you say . i have mentioned it on here before . you just have p.c. running and program and wireless speaker hooked up in or near your doll . i also use voice recognition so i can actually talk with my doll as she can answer me in whatever voice i have running . look on here for timetravelers a.i. doll . it's on here . to me this would be easier then what all you are doing . unless you just like the learning and adventure of your project.

.

philpw99 · Post by **philpw99** » Fri May 11, 2018 3:32 pm

AlexaTilbrook wrote:Nice link to Eben Upton's blog, @philpw99... but, Android isn't 100% there yet for the Broadcom VideoCore IV GPU. And The Raspberry Pi Foundation doesn't offer any official Raspberry Pi armhf builds. (Source: https://www.raspberrypi.org/forums/view ... p?t=205912 [1]) LineageOS (fka Cyanogenmod) does offer an Android Nougat (7.0) build, albeit unofficial. Google "lineageos raspberry pi 3 site:xda-developers.com" for downloads.

Sorry that I don't have enough knowledge about Raspberry Pi, and I think my skill set is too limited for programming it (I'm a python moron). So my app won't go to the Pi for the near future, sorry!

timetraveler1 wrote:what you are talking about with voice and custom talk etc. can all be done with the ultra hal a.i. program (pc) i use it and have a wireless speaker near my angela doll , you can create brain plug ins as i did for words that trigger custom conversation . i have done that with x rated talk . write anything that i want my angela doll to say . plus being a a.i. program she actually learns from things you say . i have mentioned it on here before . you just have p.c. running and program and wireless speaker hooked up in or near your doll . i also use voice recognition so i can actually talk with my doll as she can answer me in whatever voice i have running . look on here for timetravelers a.i. doll . it's on here . to me this would be easier then what all you are doing . unless you just like the learning and adventure of your project. .

I am aware that some PC programs have fully developed voice functions. However, a PC is not for everyone, there are things not easily done by the PC. Android has the best hardware platform, mobility and features IMHO. And almost unbelievably no similar Apps are offered for doll talking! I admire your work, it's marvelous, but I believe more in Android system, and also like a little control over my source-code.

===============

Now it's a little update on my project. There is none!

I am currently engaging in compiling an ultimate dolls database for myself. Like I said before, I am good at filemaker. The doll database will help me a lot to learn about the doll business. So far I had put all the WM, DS, 6Ye, Climax, D4E, DH168, OR, Piper, and YL dolls information in my little filemaker database. Currently working with the Irontech. I love the progress so far.

It's interesting to see all those chaotic data to form relationships and go to their own places.

I am not a good multi-tasking person. My brain only works on one thing and hard to switch to another. Therefore when doing one task I will halt another. The app will have to wait for a while when I compiling this ultimate list of dolls. I will update here once I have new version of the app. Sorry about the delay!!

Begog · Post by **Begog** » Sat May 12, 2018 10:22 am

timetraveler1 wrote:what you are talking about with voice and custom talk etc. can all be done with the ultra hal a.i. program (pc) i use it and have a wireless speaker near my angela doll , you can create brain plug ins as i did for words that trigger custom conversation . i have done that with x rated talk . write anything that i want my angela doll to say . plus being a a.i. program she actually learns from things you say . i have mentioned it on here before . you just have p.c. running and program and wireless speaker hooked up in or near your doll . i also use voice recognition so i can actually talk with my doll as she can answer me in whatever voice i have running . look on here for timetravelers a.i. doll . it's on here . to me this would be easier then what all you are doing . unless you just like the learning and adventure of your project. .

Let him roll, timetraveler. What he is creating is different than Hal, and the Kari program I have. Android, especially it's VR, is superior to WinVR, you don't have to buy Dragonspeak or the Hal program, it's not on the cloud, so the delay is gone and privacy is maintained, and it's completely programmable unlike Harmony (which also uses Android because of these advantages). Sure it's got a long way to go, but it is new and showing promise. Please don't anyone discourage what philpw99 is trying to accomplish here. I believe he is headed in the right direction.

philpw99 · Post by **philpw99** » Sat May 12, 2018 3:23 pm

Begog wrote:
timetraveler1 wrote:what you are talking about with voice and custom talk etc. can all be done with the ultra hal a.i. program (pc) i use it and have a wireless speaker near my angela doll , you can create brain plug ins as i did for words that trigger custom conversation . i have done that with x rated talk . write anything that i want my angela doll to say . plus being a a.i. program she actually learns from things you say . i have mentioned it on here before . you just have p.c. running and program and wireless speaker hooked up in or near your doll . i also use voice recognition so i can actually talk with my doll as she can answer me in whatever voice i have running . look on here for timetravelers a.i. doll . it's on here . to me this would be easier then what all you are doing . unless you just like the learning and adventure of your project. .
Let him roll, timetraveler. What he is creating is different than Hal, and the Kari program I have. Android, especially it's VR, is superior to WinVR, you don't have to buy Dragonspeak or the Hal program, it's not on the cloud, so the delay is gone and privacy is maintained, and it's completely programmable unlike Harmony (which also uses Android because of these advantages). Sure it's got a long way to go, but it is new and showing promise. Please don't anyone discourage what philpw99 is trying to accomplish here. I believe he is headed in the right direction.

Oh, thanks Begog! That's very kind of you! Thanks for your concern! I am actually a person easy to get discouraged, but I hope this app will not suffer this fate like my other failed projects. I really like it myself, and I also wish Debbie will talk to me like a proper lady soon.

Ninja Gaiden · Post by **Ninja Gaiden** » Sun Jun 24, 2018 4:56 pm

Any updates?

philpw99 · Post by **philpw99** » Sun Jun 24, 2018 9:33 pm

Oh, sorry! So far no updates. 2 of my relatives came and they will live in my apartment for 3 months. They are sleeping in my room, and I am sleeping in the sofa. Since I can only do the programing in my room (can't do it in other places, I tried), there were no progress I can report to you all, hence the long silence. They will go back to China next month, only then I can start pick up the work again.

Just a little extra, now I am also thinking about implementing wit.ai in the future program, or some kind of naughty AI which can handle the natural language, but since it's just a thought, and the wavenet is just started (for better voice), I already have a lot in my todo list.

Also, I am planning to go back to China in Sept. So at that time there won't be any update also. Just want to say it before hand, unlike this time.

Begog · Post by **Begog** » Mon Jun 25, 2018 7:42 am

It's good to know this project isn't dead. Please work on the speech programming first. If you get that fixed, I'll have something to work with and go back to programming an interesting profile or two.

philpw99 · Post by **philpw99** » Tue Jun 26, 2018 7:26 pm

Begog wrote:It's good to know this project isn't dead. Please work on the speech programming first. If you get that fixed, I'll have something to work with and go back to programming an interesting profile or two.

Will do that. The next version is all about the fixing problems and enhance my filemaker profile making program. It will be much easier to make profiles with that filemaker thing.

AlexaTilbrook · Post by **AlexaTilbrook** » Thu Jul 05, 2018 2:17 pm

A little quick reply...

I have been following this topic and I have been interested in it, I also have my amateur radio license now... And I have a Google Home Mini.

I uploaded two videos of example IFTTT (If This Then That) interactions...

I'm going to ask philipw99 if something like this is possible...

And the IFTTT "programming" is so trivial anyone can do it...

: Screenshot_2018-07-05-12-44-38.png (97.1 KiB) Viewed 2066 times

That something like this could be implemented in the Doll33 APK?

I bought the Google Home Mini as an introduction to artificial intelligence. I already have her controlling the thermostat (a Nest smart thermostat), waking me up, and stuff I wish Daphne could do. Now only if I can rename my Google Home Mini to answer to Daphne... Now that would be awesome.

I just thought I'd post this as an encouragement to the OP.

philpw99 · Post by **philpw99** » Thu Jul 05, 2018 8:15 pm

@AlexaTilBrook

So far it's only possible to launch the Google assistant, but not sending commands to it. So I can implement something like "Hey, can I ask google something?", then my app will say, "Of course, any time." then launch the google assistant.

There might be some way to call the assistant api and send the command, like this guy says:
https://stackoverflow.com/questions/468 ... ndroid-app
but I am not sure his implement will work, since he actually just sent a text search, not a command.

A better way is to call the IFTTT directly using their API:
https://github.com/IFTTT/IFTTTSDK-Android
This can be done in a much direct way. However, the voice is recognized by the phone, not Google API, so the accuracy is not as good as the Google Assistant.

Btw, my Doll33 app is using phone recognition, so the accuracy suffers too.

@ Everyone
Recently I have been giving my app a lot of thought, and thinking which way it should go in the future. Since I cannot do the programming until my uncle and aunt are gone, I have plenty of time to reflect on my plans.

The more I think about it, the more I am leaning toward a "crowd sourcing" approach. It's inspired by the Wikipedia, crowd sourcing can pool together everyone's effort, and the data accumulation and improvement will be much faster than the regular approach.

Imagine that you run the Doll33, you ask, "Do you like me, baby?" Of course the doll will answer "Yes, I do." If she was programmed to do so, but you can improve upon the simple answer to something like, "Of course I like you the way you are, baby." You can just type the answer yourself. And if someone already entered that answer, you can rate this answer from 1 to 5, or "like" and "dislike". So when more people ask the same question, the Doll33 app will have more answers, and it will choose the best answers to reply you.

In this way, there is no need to do any chatbot programming. The online database will simply pool everyone's effort together and give you the answer most people like. Also you can create your own answers or conversations right inside the app, and your answer will help the app answers better. I believe this way we will have a nice doll conversation database in a very short time.

The downside of the approach, of course, is that the app had to be online in order to do this. Your questions and answers(without your private information) will be sent over to the online database. And if someone purposely put those bad sentences (racial or hate based), he might be able to do it. Microsoft had a bad experience with that. But I think the people in the doll community will not do such things.

What do you think about this crowd source approach? I know you all like the offline database, but please give this approach a thought. Thank you!

gator048 · Post by **gator048** » Thu Jul 05, 2018 8:59 pm

: 20180705_134147_1530814888819.jpg (4.13 MiB) Viewed 2048 times

Make it easy on your self and get the robotics by realdoll

Begog · Post by **Begog** » Fri Jul 06, 2018 9:08 am

There are several things I don't like about this approach. #1) It uses the internet now. That significantly slows it down, opens you up to hackers, severely limits privacy, and causes it to not work when out of range or weak signal. 2) Group answer? So much for programming individual "personalities." Now everyone's avatar will be exactly the same, much like Harmony. Might as well use Harmony. This is not the direction I hoped you would go in. Good luck with your project.

philpw99 · Post by **philpw99** » Fri Jul 06, 2018 1:00 pm

gator048 wrote:Make it easy on your self and get the robotics by realdoll

I want to make this app works for all the dolls, not just single manufacture one, and you can use all kind of android phone or equipment to implement the way you want. So it's very different. A good comparison is Apple's OSx and generic Windows. OS X works only in Apple, but Windows works for the rest of the computers.

Begog wrote:There are several things I don't like about this approach. #1) It uses the internet now. That significantly slows it down, opens you up to hackers, severely limits privacy, and causes it to not work when out of range or weak signal. 2) Group answer? So much for programming individual "personalities." Now everyone's avatar will be exactly the same, much like Harmony. Might as well use Harmony. This is not the direction I hoped you would go in. Good luck with your project.

I knew you wouldn't like it, but I am actually thinking about adding an online mode, in addition to the current offline mode. The Doll33 app already has a rudimentary conversation script system. Fixing it and make it work better is not that hard, and you can use Filemaker to make proper profiles soon. However, I am thinking about the next step, the online mode, which you can turn on and make the App much smarter, and speaks much more natural. The online profile can have different personalities as well. I think "innocent girl", and "naughty spicy" are the first 2 profiles I will put on.

The offline mode will still be there, and I hope I can even provide the feature of downloading the whole online database, to use it offline, which I hope it will be appealing to you.

However, in order to have a decent size of conversational database(over 1mb of data), online co-operation with everyone using the app seems the best way to do it. Begog, you are a very good chatbot programmer, but how many lines can you put in a chatbot, and how many different way can you answer the same question without killing yourself in the process?

Just like you said, Harmony is not customizable, but Doll33 will be created by us, and customized by us. Only thing I control is how many "personalities" there will be in the big database.

And I forgot answering your points, here they are:
1) It will only send text and downloading wavenet voice, which use little bandwidth. Even it might be one second or two slower than the offline mode, in an conversation we wouldn't mind much.
Hackers will have a hard time hacking google. Their servers are rock solid. The app will never gather the popularity for hackers' effort anyway.
Privacy concern is only about the things you say to the App, and your contributed answers. It's the same concern of making a chatbot online.
I suppose most doll owners use the App indoors, which has strong Wifi signal, and you can use the offline mode when signal is weak.
2) Group answers.
Everyone will benefit the same effort and they can participate in making their favorite profiles better. That's my view of the group profiles. I don't want to make one profile just for one person(doll).
I hope when contributing the answers people will follow the doll's personalities. I am not sure if people will follow that rule, but I sincerely hope so. The rating system might be the way to prevent some nasty people from sabotaging it.
3) The direction of the app eventually will be online. There are just too much benefit and advantage of using Google's powerful AI platform. Wavenet voice, naturally language processing and online databases, to name a few, are very good for a doll talking program to improve upon. I don't want it to be the second Harmony, but I do want my App to go beyond just prepared profiles.

philpw99 · Post by **philpw99** » Sat Jul 21, 2018 12:53 pm

Alright!! Everyone, here is a little update: my uncle and aunt finally went back to China, after 3 months staying in my home. It felt like a Decade! Finally I am back to my bedroom and I can code again! Hooray!

Thank you everyone for keeping an eye on my little project! I hope I will not disappoint you guys further.

The first thing I am gonna do, is to separate the Doll33 app into 2 apps. It's proved that Google will not tolerate anything sexual in the play store. And I don't want to simply cut out the F-Mode from my app. Plus the App need some major clean up any way. (It is my first mobile app, so many things inside are messy.)Therefore I will make 2 apps. Both will be built almost from the ground up.

The first one I will make is called "Doll33Love". It will be for TDF members only. Its only function is the F-Mode. There will be different profiles and sounds packages. I will make the profiles using my own Filemaker programs, which I will share with the community. Or you can make your own profiles using an text editor, then zip all files together. The app itself will not have a way to change the profile and sounds. It will save me a lot of time in programming and keep the app small and efficient.

Also I will try to find a way to "listen" when the F-Mode starts, like if you say "I am touching your ...", it's the same or even stronger than the movement of up and down. Hopefully you can use only your words to make the doll feel high.

This app will also have "sleeping sound" feature. Once you click a button, the app will play some sound in an endless loop. And some of the sound are randomly made, like people do when they are sleeping.

After this app is done, the next one is Doll33Talk. As you can guess, this app will be available in Google Play Store and it will only do the talking. Or maybe more, like turn on your TV or lights using google assistant/IFTTT. I am planning it to have both offline and online profiles. Just like the Love app, you cannot change the profile directly in the app, but only with my filemaker app, or your favorite text editor. The offline feature will be implement first, then the online profile will be my long term goal. IMHO online database is the only way that can make the app smart.

anachorn · Post by **anachorn** » Thu Aug 23, 2018 7:44 am

Great Idea, fairly simple of an app to make as well. I was looking on her to see if someone was looking into something like this. I'll download it for sure if one is made. Maybe it could be a good starter app for myself to do. I am a self employed game developer, and I have created lots of little apps for iPhones for myself to play. Could dabble in this and some other app ideas if interested. PM me if you are doing something like this as well. Maybe we could join together and pump an app out for everyone ^_^

philpw99 wrote:I just had an idea of putting a miniature Bluetooth speaker, like this one from Amazon, in my doll's mouth. Then make an app in android to enable her to speak.

I saw the VoiceAttack program people talked about here, but it seems too limited. I want my app be able to do a lot more, like:
* In the morning she will say "Good morning Philip, time to wake up." at my wakeup time.
* She will say "Good to see you back!" or some similar random sentences when I start the app.
* Whenever she hears "love" or "like", she will say something lovely back at random.
* The most important thing: I will put the phone on her and start the "F" mode, she will feel the shaking by the phone's accelerometer , a "heating" number will start to accumulate, from 0 to 100 she will say "stop" or "don't touch me", 100 to 500 she will start low moaning, over 500 she will moan in full volume. Then when the number drops she will say "don't stop"... etc. You got the idea.

I will use opensourced NativeScript for this app, so in theory both Android and iOS can run, but I only use Android to test. I will keep this project free and open as long as I can.

Also it will have profiles, so you can switch her to speak in different languages.

What do you think about this project? Please give me some ideas and suggestions. Thank you!!
(Noted) I haven't written any code for it yet. Still in early brainstorming stage, but I will write it for my incoming doll. Hopefully it will benefit everyone here too.