Starting a code project to make my doll speak

[LovesBlackWomen]

Wouldn't it be easier just to stick a tape player in her head? I've done that since the 1970's.

[AlexaTilbrook]

The phone's accelerometer? What's that?

This. Basically the thing that makes your screen rotate from portrait (tall) to landscape (wide) and vice versa, just by turning the phone around.

I support a stand alone app. Love the idea of a speaker in her mouth. I'd like the ability to upload my own script of what she says, as well as in the F-Mode to adjust the parameters of sensitivity. Some of us (me) may never be able to afford an animated head. You know what? My imagination works just fine, and your idea would be most welcomed.

Seconded!

[philpw99]

Also for each phrase the app will say it with a realistic voice (maybe recorded mp3?), or if it cannot find the voice file, it will speak with built-in text-to-speech engine, which will sound like a robot.

Another nice thing about VA is you can use any high quality SAPI5 voice like the ones from Ivona as well as recorded mp3s.
Their British English Amy is one of the best I've heard.
Here's some TTS with Amy...

introduction.mp3

I'm not sure if this could give you some more ideas, but you can check out my audio interactive Eva mannequin project here: viewtopic.php?f=140&t=69149

VA is not lack at sound quality, it's missing an accelerometer. The sound quality for most people is not that big deal between lossless FLAC and mp3s. But people always can tell the robotic feel in a TTS speech. A lot of progress had been made with the TTS, but still not good enough for a doll's voice, I guess. It will be there only for people who cannot record a girl's voice, but still want the doll to say some particular things.

Your Eva project is a great attempt. It does give me some ideas, but not much with this voice project though.

oneatatime, thank you for your tips. It's my favorite website to get free sound too. I previously worked with a telephone project, and I got some nice sound from there.

Anung Un Rama and Boobman , I love Genie too, but not really getting hard for her. She is just very funny to me. It will be up to you guys to make the genie voice profile.

LovesBlackWomen, hahaha, you are really old school. Today it's wearing an VR helmet with a doll on top. No more tapes.

[Huckleberry]

I think voice is the great leap towards a true love doll. After all, human companionship is based on conversation for the most part. Even if a pre-programmed loop of noises, moans, squeaks - all those little personality traits we love - could be played while we get intimate, that would satisfy me until the day I have my very own walking talking robotic doll.

I use the girls voice changer app for cosplay, and allied with the small BT speaker in her mouth, it could be the nearest experience. If they could invent a smaller moving speaker that worked in tandem with the sound, it could look like she was actually speaking.

[MannequinFan]

VA is not lack at sound quality, it's missing an accelerometer.

Not sure if you've looked into it, but I think Fleshlight's Vstroker uses this type of technology.
I've never used it, but it seems to be designed to do pretty much what you are describing.
Supposedly it even works with sites like 3d Sexvilla where you can create your own custom scripted scenes.
I would assume that would also include custom audio tracks. Looks like it could be fun to play around with.
https://www.3dsexvilla.com/customization/

Not trying to divert you from your project, just some things to look at that are already out there.

[oneatatime]

Great idea. Did you get sample voices?

yes I was able to download a number of samples. I used audacity to edit a number of sounds together. Then use the my phone and a small Bluetooth speaker near by to get some excellent phrases and sounds.

[TDFnoob]

Just a thought, but perhaps doing a "brain implant" and inserting something like an amazon echo / "Alexa" would work? You could bore through the nasal and ear holes pretty easy that way. It's already got most of the personal assistant, "wake me up" type stuff installed, and they're working on a decent chatbot that can talk for a few minutes about a ton of topics.. They have a few different "accents" avail, it may be that changing a voice is possible too..

Admittedly, I don't know what's involved in that. Heck, I just bought my first doll, which is supposed to be here Wednesday, but this is a topic i've got a SERIOUS interest in.. I was actually thinking of posting my own suggestion thread...

[philpw99]

Huckleberry,
Yeah, I have the same idea of the moving speaker. I will be in China this September, and I will ask around, and see anyone there interested in making things like that.

Mannequinfan,
I have never seen Vstroker before. It's the right idea, but used only in their own devices. I like an open platform which allow any smart phone and any doll more.
The interaction of 3D sexvilla is something I want to achieve in the future. I prefer normal conversation over sex talk though. So I like the doll just say "how are you" instead of "I am horny". So it's better for the user to edit the dialogue themselves.

oneatatime,
Great to know what you did, maybe I can borrow your sound files in the future?

TDFnoob,
Actually OK Google is the thing I want to implement, even in the early versions. You can just say " <doll name>, ask Google for the weather", then the request will be passed to Google. I will check if I can pass the question to Alexa. It shouldn't be too hard.


Now here is a big question for everyone here (if you are still following this) : Do you want to have an online profile for your doll's profile, or you prefer just the local database?

Online doll profile (synchronize with your phone):
Pro: * You can use different phone for the same doll. * You can edit your doll profile online through a web browser. Editing your conversations and events, adding voice files will be much easier through a browser. * You won't lose your doll profile, even your phone is destroyed.
Con: * You will worry about exposing your privacy * You need internet to change your profile. * If my website got hacked, bad things could happen. (cross my fingers)

Offline profile:
Pro: * It runs on your phone only. So your privacy is safe. * You can edit the profile directly from your phone, thought it might be hard (screen too small).
Con: * You will lose all the settings and conversations if you lost your phone, along with all your sound files. You might need to spend a long time to rebuild the profile. * It's more difficult to transfer the voice files to your phone and assign them to your conversation or F-Mode.

So what do you think guys? Online sync or offline only? I like to know which approach I should use in the beginning to determine the way I program it. Thanks for any input!!

[AlexaTilbrook]

Online doll profile (synchronize with your phone):
Pro: * You can use different phone for the same doll. * You can edit your doll profile online through a web browser. Editing your conversations and events, adding voice files will be much easier through a browser. * You won't lose your doll profile, even your phone is destroyed.
Con: * You will worry about exposing your privacy * You need internet to change your profile. * If my website got hacked, bad things could happen. (cross my fingers)

Offline profile:
Pro: * It runs on your phone only. So your privacy is safe. * You can edit the profile directly from your phone, thought it might be hard (screen too small).
Con: * You will lose all the settings and conversations if you lost your phone, along with all your sound files. You might need to spend a long time to rebuild the profile. * It's more difficult to transfer the voice files to your phone and assign them to your conversation or F-Mode.

So what do you think guys? Online sync or offline only? I like to know which approach I should use in the beginning to determine the way I program it. Thanks for any input!!

Two things:

Regarding profiles:

I prefer the hybrid approach. Some people have soft-capped data quotas on their mobile plans and sometimes wifi isn't an option for them. Some people possibly would want to use a device that had been deactivated (due to whatever reason, like phone number porting, device upgrades, etc.), or want to finally put to use that El Cheapo no-name Android tablet Aunt Donna (just some random name I thought up) got them for Christmas from the Kohl's Black Friday sale... the "android.permission.INTERNET" permission is a MUST. It can be used to sync profiles between server and your app (client), but if internet is not available, then a Toast notification should be shown saying "Network unavailable. Your doll's profile, <DOLLS_NAME> will be in a limited mode." Or something like that.
And Android can be run on the Raspberry Pi... its kinda difficult to set up and the Broadcom VideoCore GPU isn't supported well in the Android kernel, but its usable, albeit slow...

That being said,

Would you mind if I ask to assist on development? My doll Elisabeth has her own GitHub account we can use for development collaboration (https://github.com/LissieTilbrook), and I know the Android system inside and out. PM me if you're interested.

[AlexaTilbrook]

I should also say, the profile sync could be an option...

Free app: Ad-supported (Google AdMob), but a connection to the Internet is a requirement. No network, app no worky.

Paid version: No ads, the option for offline profile storage and maybe the app can work without Internet. Can be either an in-app purchase or a separate download on Google Play.

I believe people would like to probably want to first download a free version of your app just to see if it really works. The wide diversity in Android devices and minimum SDK versions can throw the proverbial wrench into your app's retaining rate (and yes, Google Play does keep track of that), due to users leaving 1-star reviews complaining of app not working, app is crashing on their phone, it throws an ANR (App Not Responding)/FC's (Force Closes back to launcher with, sometimes without, a notification [but Android Nougat 7.x has made the latter infrequent lately]) on them, or the rare XDA'er or someone similar complaining about the "Byzantine amount of permissions requested--do I want my chatbot really having access to my contacts?" or some sort of nonsense...

But if you state in your app's description that you would prefer that if issues need to be raised, email you directly, create a topic here (on TDF and ask a staff member to make it sticky), and link said support topic on your app's description; that any and all negative reviews with "it just crashes" will be ignored... then your app might have a user base. But people would just ignore your request to direct support requests to another channel, they will just lament on how "it's a piece of garbage, uninstalled" (that's just an example of what could be said, some have wording I refuse to repeat here), anyway. If that happens, reply to the person who left such a review. Maybe they'll read it and offer more info. If not, forget about that person. Also if someone leaves a five star review without any reason, reply to that review too. Again, they would probably want to edit their review to explain, but be careful, because some of these "blank five-star reviews" are probable spam reviews done by people to commit drive by download and review and then delete acts just because some app that claims "do stuff from your phone and get paid--download and rate apps, watch videos, read websites and accumulate points/whatever for Amazon/Google Play/PayPal gift cards or CASH!"... beware of those.

But if real users really like your app, then they'll WANT to pay for it. If they like it, they want to have extra features, or just "to support the developer," or whatever reason, including to get rid of ads, you have to give your users an incentive why the paid version is worth it. If a paid version is just to remove Google AdMob code, people won't see any reason to pay $1.99 or whatever. Introduce a taste to them to see if they would use it. If they like it or it is just another same old thing some have seen already on the Play Store.

And importantly, if it works with their phone. What might work on a flagship Samsung or LG might be a crash-fest on a lower end Samsung or LG, like the ones sold by some prepaid carriers. If it does on their phone, and would want the extra features, they'll pay for it. If they're just satisfied with the free version, because it's just sufficient enough for them, they'll keep the free version installed. Maybe they'll buy the paid version someday, maybe not. They're using your app. On the other end, if they find out it won't work, they'll uninstall it. They might leave a 1-star review saying why, but keep up on those. But the ones who got hooked on your app will probably pay for it.

Again, Philip, I'm extending the olive branch for you. You touched upon a topic near and dear to me, and has been an interest for almost all of my life. I'll be glad to help you on this.

[MannequinFan]

Here's something else I stumbled upon a while back...
https://dialogflow.com/
It used to be called API.AI.

Not sure if it would be helpful, but it could be a starting framework for building a conversational bot.
It does offer "hooks" to APIs for getting time, weather and that kind of thing and can apparently be used with assistants like Google Home and Alexa and the like. There is also a pre-built "small talk" function that can be edited and added to.

[oneatatime]

Huckleberry,

oneatatime,
Great to know what you did, maybe I can borrow your sound files in the future?


Now here is a big question for everyone here (if you are still following this) : Do you want to have an online profile for your doll's profile, or you prefer just the local database?

Online doll profile (synchronize with your phone):
Pro: * You can use different phone for the same doll. * You can edit your doll profile online through a web browser. Editing your conversations and events, adding voice files will be much easier through a browser. * You won't lose your doll profile, even your phone is destroyed.
Con: * You will worry about exposing your privacy * You need internet to change your profile. * If my website got hacked, bad things could happen. (cross my fingers)

Offline profile:
Pro: * It runs on your phone only. So your privacy is safe. * You can edit the profile directly from your phone, thought it might be hard (screen too small).
Con: * You will lose all the settings and conversations if you lost your phone, along with all your sound files. You might need to spend a long time to rebuild the profile. * It's more difficult to transfer the voice files to your phone and assign them to your conversation or F-Mode.

So what do you think guys? Online sync or offline only? I like to know which approach I should use in the beginning to determine the way I program it. Thanks for any input!!

Hi

I'd be happy to send some of my sampled sounds

wrt online vs offline

I'd vote for offline, thanks!

[philpw99]

Thank you everyone for your great replies!! Really appreciate them!

AlexaTilBrook,

Of course it's great to have an expert like you in a team. I think you will greatly accelerate the process, and thank you so much for your advice! However, I am not an expert programmer like you. I did mostly Access, Ms SQL, filemaker, autoit programming and some visual basic ones. I am still learning Typescript and Javascript at this moment. This project is a way for me to learn mobile programming. I don't think I will make much progress in the beginning, so the first primitive version will be very bad. It will also rely heavily on other's plugins to work. Therefore I am not sure if you can bear a new beginner like me. Also I probably will keep it a private project until I can have something presentable, only after then I will bring my little pet project to you, and we will all make it better.

I like to keep the program free of ads. So my price plan is all about the voice: free version will only have the TTS voice, and you can edit the text anyway you want. Then you can choose the "Premium" profiles to purchase. Premium profile contains all the personalize conversation and the voice recordings. Those are the ones really time and money consuming, therefore they cannot be free. I believe once you use the free version, modify the text and understand what the program can do, you will buy the premium profiles in a heartbeat (presume that I can actually make them).

F-Mode is a very interesting thing to think about, I think the free profile will have F-Mode, but you need to pay a little to enable it. Not because I want money, but just want to make sure children will not get it by accident. But of course in the beginning months all features are free for all.

Moreover, for me I think doing offline database IS the easiest way to go, and the "premium profiles" will be the real money maker in the future. The app itself won't make much money.

Mannequinfan,
I saw that website. It's a great conversation approach. Too bad it's not for my little pet project. I don't need the app to be that smart, but I want it to have different scriptable personality and voices. Dialogflow cannot do that. It can be an assistance, like the google and alexa, but not the main engine.

oneatatime,
Cool... yeah, offline is better for you, and easier for me.

[TDFnoob]

I think an online profile would be better for you, and more financially viable.. Face it, it's a lot easier to control access when people have to go to you..

Now, advantages for your end users. It's going to be a lot easier for them to preview stuff... Click a button, hear a sound font. Or heck, you could even set it up to where people could type stuff in and have it parroted back to them so they get a REAL demo of how things will sound with the particular voice/profile.

That being said, why would you use the same password for everything? So, someone hacks your site, do you really think I used the same password anywhere else? Sure, there's always going to be a couple people that do. But the ones that are SERIOUSLY concerned about privacy aren't going to put their information out there to begin with.

One thing, I can't remember if I read it on this thread or another, but if you're using the phone to run the voice for the doll, have you considered perhaps using the acceleromators in the phone? Could be used to detect repeated motions for moaning/etc?

[philpw99]

I think an online profile would be better for you, and more financially viable.. Face it, it's a lot easier to control access when people have to go to you..

Now, advantages for your end users. It's going to be a lot easier for them to preview stuff... Click a button, hear a sound font. Or heck, you could even set it up to where people could type stuff in and have it parroted back to them so they get a REAL demo of how things will sound with the particular voice/profile.

That being said, why would you use the same password for everything? So, someone hacks your site, do you really think I used the same password anywhere else? Sure, there's always going to be a couple people that do. But the ones that are SERIOUSLY concerned about privacy aren't going to put their information out there to begin with.

One thing, I can't remember if I read it on this thread or another, but if you're using the phone to run the voice for the doll, have you considered perhaps using the acceleromators in the phone? Could be used to detect repeated motions for moaning/etc?

Online profile has the advantages of what you said, I completely agree. Then it will require me of double programming: online, offline and sync. There are plug-ins that make it easy, but still it will require many more hours of testing and maintaining. I prefer just focus on offline. You can click on a button to hear how the conversation sound anyway.

I don't actually want to make it too easy for my users. I like them to just buy the premium profiles and be done with it.

Right now most of the planning is done. Here are some interesting things of this app:
* I will use an variable indicator called "Love Point". It's like HP but it's for how much the doll loves you, or hates you. LP 0 is "neutral", she treats you like a friend only. LP above 50 is "Like", she is in favor of you. Above 100 is "Love". She loves you. Below -50 is "hate". She will talk trash to you. Different level has different dialogs. If you say "good morning, dear.", she will reply according to the LP. If she hates you, she will say "go to hell", "leave me alone". If she loves you, she will say "I love you too, dear." or "Oh, how sweet you are..."
* If you say "I love you", LP will increase 5. If you say "bitch", LP will decrease 5... You get the idea. Every day LP will get closer to 0. So if you don't say "I Love you" for a few days, she will return to being friends again. Also you cannot say the same thing over and over, she knows you already said it.
* All dialogs are randomized, for the same question, she will answer randomly in a list. Like the above example, she might have 5 replies to the same "good morning" question.
* In F-Mode, which will definitely use the accelerometers, the doll has a HP (Heat Point), it always starts with 0, then when it detects a motion, the HP will increase, so the moaning sound will getting louder and louder. When the HP reach 100, the volume will be the loudest, and the sound files will be different. She will start to yell, scream or say some nasty things.

However, please don't get excited about the app yet. It's still in very early stage. The planning means nothing when it comes to actual programming.