SPEAKER AREA IN DOLLS HEADS FROM MANUFACTURER

MannequinFan · Post by **MannequinFan** » Mon May 04, 2015 11:52 pm

Begog wrote:The problem I have using the webcam mic is that it "hears" verbot talking and answers itself.

I don't seem to be having that problem with HAL since I turned off the Windows voice recognition - HAL doesn't need to have it running.
I can see the sound level meter going when I talk, but when HAL talks there is no meter activity. Apparently it knows not to listen when it's speaking.
Not sure if Verbot or Kari can use the recognition without it actually running.

I guess what is needed is an "auto mute" function that would mute the mic when the computer is playing sound.

timetraveler1 · Post by **timetraveler1** » Tue May 05, 2015 9:59 am

MannequinFan, i saw your question , i think a person has to go into the brain editing , you have to be careful though , as just with any program what you do can affect other things , such as blank hal responses ( i have had that in the past as others have to . )
hal a.i. has great capabilities though .
now people know why i wasn't just going to give my a.i. brain away , i have done many custom things to it and have years (over 6 ? years ) of learning etc. tied up in it .
yes hal knows when to listen and not talk when you are talking with hal.
I do wish that manufacturers would create a speaker spot in the dolls head like the one you? printed on your printing machine .
if you turn off windows voice recognition you will need another vr to use with hal , otherwise you have to just type . and that wouldn't be any fun !

If anyone is ever writing to me on here and i don't answer for a while , you can also pm me ,( i will write you back ) sometimes i don't check this area all the time . as i check many area's on the forum .

Begog · Post by **Begog** » Tue May 05, 2015 6:17 pm

MannequinFan wrote:
Begog wrote:The problem I have using the webcam mic is that it "hears" verbot talking and answers itself.
I don't seem to be having that problem with HAL since I turned off the Windows voice recognition - HAL doesn't need to have it running.
I can see the sound level meter going when I talk, but when HAL talks there is no meter activity. Apparently it knows not to listen when it's speaking.
Not sure if Verbot or Kari can use the recognition without it actually running.

I guess what is needed is an "auto mute" function that would mute the mic when the computer is playing sound.

The webcam mic is pretty useless for that reason, and the BT headset works better. However, I see the problem with using Windows VR, because it seems to have a small vocabulary, especially when it comes to "dirty talk." I will say something nasty and it will auto-correct it into something benign that makes no sense and ruins the conversation and mood. Does Dragon do that too? I may have to buy it. And what about Karigirl? Does she need VR too? The vid I watched the guy was typing. When I select "Voice Recognition" in Verbot, Windows VR cranks up.

Augustus15 · Post by **Augustus15** » Wed May 06, 2015 8:37 am

I sent in a help request to Dragon, and it seems that the basic commands are set in stone and cannot be altered.
Dragon is quite expensive but i am using it with Kari as a secondary function, i mainly use it for my work related emails and texts, and for that it is invaluable and cuts my work time by about 50%.
With Kari pro you have to work on the "What i say, what you say" but you need to put in a lot of alternative questions as you are likely to forget your original ones, also you can add to, modify, and clean up the personality file that you have created.

timetraveler1 · Post by **timetraveler1** » Wed May 06, 2015 2:18 pm

Augustus15 i only paid about 35 dollars? for my dragon speak on ebay it was a 10.5? and upgradable to 11? for free .
i don't really know if you can talk dirty to it or not , i never really tried lol . but i imagine it will type to say anything . ms windows vr is designed probably for family .

Begog · Post by **Begog** » Wed May 06, 2015 3:57 pm

So I guess the answer is "no," Karigirl needs Win VR or Dragon as well. Like Karigirl, and I guess Hal, Verbot needs programming to respond the way you want("What i say, what you say"). Verbot runs perfectly on my old XP PC which can't seem to install BT software, even with a dongle. It apparently has a crude version of Win VR, but i haven't tried it. On my Win7 laptop, it talks too fast. So fast it skips some words and fails to pronounce others correctly. Add that to Win VR failure to type in what i said correctly, and the conversation goes to hell in a handbag. All that stops when I go back to typing except the talk speed, and I see no way of slowing that down. I don't think it is going to work properly with a modern computer, although you should hear this thing blabbering with the TV on. Response time is fast.

MannequinFan · Post by **MannequinFan** » Wed May 06, 2015 4:13 pm

You'll have to teach the MS VR all the dirty words. Start speech redognition and right click on the microphone icon in the listening box. Go to Open the Speech Dictionary and go to Add a new word. You can also say the word in dictation and correct whatever the computer comes up with by spelling the word you want. It will ask you whether you want to save the new word. This might be better as it will learn the pronunciation also.

I think I finally found the perfect program for this (for me anyway).

Thanks to member perciousnow for posting about it a few pages back.

If you aren't wanting to spend a ton of time trying to train an A.I. program, Voice Attack is a fantastic little piece of software. I've been messing with it today and it has some really cool features and is super cheap at only 8 bucks to buy it.
http://www.voiceattack.com/
It's made for voice gaming control and works seamlessly with the MS VR components (MS VR doesn't need to be running) and seems to be very lightweight compared to HAL. It's not an A.I. program but the really, really cool thing about it is that you can create responses that can contain both played sound bytes and TTS in the same response. So you could, for instance, have her giggle (sound byte) and say something (TTS). So I can use the Acapela recordings along with the Amy voice to create responses to my spoken commands. Very cool. I can see having hours of fun with this!

The interface is very simple - just create a profile for your doll and start adding voice commands. The trial version has only one profile and 20 commands, but the full version is unlimited.
By far the coolest feature is how you can stack any number of responses in the same command. So you can have it play a sound byte, then say someting, another sound byte and so on.

: command stacking.jpg (196.47 KiB) Viewed 3897 times

The "other" button is where all the goodies are.

: other stuff1.jpg (178.14 KiB) Viewed 3897 times

If you put several responses in separated by semi colons, the computer will randomly choose one as the response.

I can see almost endless possibilities here.

: other stuff.jpg (148.9 KiB) Viewed 3897 times

One other really nice feature is that you can set it so it only listens when you say a keyword, such as your doll's name.
This eliminates the problem of it hearing itself or any other noises in the background that might cause an unwanted response.
Just turn off listening and she will only listen when spoken to. I've got my webcam mic set on full volume/full quality and I can give commands from 10 feet away and be understood.
VA recommends going through the MS VR voice training 2 or 3 times which does help a lot.

: listening override.jpg (172.01 KiB) Viewed 3897 times

Augustus15 · Post by **Augustus15** » Thu May 07, 2015 4:05 am

I bought Dragon 13 from a Curry's digital store in UK a couple of months back in physical disk form. as sometimes it is a nightmare downloading stuff in China. (it's a big file, about 4GB)
You can say absolutely anything with it though, however i have also just downloaded the trial version of Voice Attack, so i will do some work with this over the weekend. Still having a job convincing John at JM Doll to put a speaker cavity in Lucy or Emma's head however, they seem to think people buy dolls so they DON'T have to have a woman speaking, yet i would buy an extra head of him if he did, mind you i told him he has to do Elf ears as well, Luna is a bit too small for me and i reckon Emma would look real sexy with Elf ears!

MannequinFan · Post by **MannequinFan** » Thu May 07, 2015 12:20 pm

Augustus15 wrote:they seem to think people buy dolls so they DON'T have to have a woman speaking

LOL, actually, I think I agree with them.
I'm really not that much into having conversations with inanimate objects myself - I think it gets a little too close to talking to myself, which just seems a little too weird for me. Just my own feeling though, not criticizing anyone else.

It would be nice to have some sexy sound effects like the heavy breathing and some "oos" and "ohs" and such that you could invoke by saying a keyword or phrase. The breathing clip that I posted earlier is really quite realistic and I think something like that could really enhance the overall experience.

I think the dead giveaway to me is that the mouth isn't moving with the talking, so the whole illusion just doesn't really work for me. However, breathing and moaning can be done without mouth movement.

MannequinFan · Post by **MannequinFan** » Thu May 07, 2015 3:52 pm

Augustus15 wrote:but you need to put in a lot of alternative questions as you are likely to forget your original ones

This is true. Same thing with Voice Attack.
In VA you can just right click on a command and duplicate it and then just change your questions a little. Very easy to make many variations quickly.
For example, you could make a command that had several responses separated by semicolons to get a random response, then duplicate that several times with slightly different questions, like "how are you doing?" "what's up", "what's going on", etc.

The difficulty with using VR for a conversation is that the computer doesn't know what you're going to say, so it just makes a guess - most of the time not quite right.
The advantage of having pre-set questions is that the computer knows exactly what to expect so matching the sound profile is much easier. The sound either matches or it doesn't.

Begog · Post by **Begog** » Thu May 07, 2015 5:58 pm

That Voice Attack looks like the ticket. I am going to have to check that out when I get some time. I don't care about anything on the screen, just what's coming out of the doll speaker, especially some nice moaning sounds on command.

Augustus15 · Post by **Augustus15** » Tue May 12, 2015 12:13 am

Yeah, checked VA out, it is certainly a useful tool and i can use it on other applications!
Lip Sync would be a great idea but i think we still have a way to go here, Full Android sexbot, more like 2050 i think. There are some great looking female androids on "Jupiter Rising" with Channing Tatum and Sean Bean.

gyxile · Post by **gyxile** » Tue May 12, 2015 7:24 am

Voice attack looks brilliant, and for $8 how can you go wrong

MannequinFan · Post by **MannequinFan** » Tue May 12, 2015 11:41 pm

I had a little fun with Voice Attack a couple of days ago.
I took some of the text from the tables in the HAL brain and created a response to the question "Eva, what are your thoughts today?".
The first part of her response was made using the text from the 3 part sentence generator in HAL.
The beginning part goes something like "I heard on the radio that...", the second part is an object like "TV repairmen" and the third part is some kind of an action that the object takes. Each of the 3 parts is randomly selected from a long list of items, so there are literally millions of possible combinations. She never says the same thing twice. Sometimes it comes out nonsense but it can be funny.
At the end is a randomly selected "Your mama is so ugly/fat/poor/stupid insult. HAL has a long list of them, some of them are pretty funny.
The Ivona British English Amy voice is just outstanding for this.

Here's a few examples...