Milovana.com

Hi everyone !

That's my first post here but I can say I've been lurking around for quite a while... ^^'

During the past days, I heard of an easy way to do chatbots, recognizing audio from the mic and responding to you, understanding quite complex sentences and beeing able to sustain a dialog without coding much... I obviously quickly though of a way of hijack that for a new kind of use... It's not an actual tease yet but just a showcase of what we could do with it.

Here it comes:
https://watson-domme.eu-gb.mybluemix.net/
(you should click on "Listen to me" after selecting your language on top of the page if you want your mistress to actually listen to you...)

It's actually built with IBM Cloud (Watson) services which are free for use at low level of usage, so I hope it will sustain your... load. ^^' I took something on github and just tweaked it a bit to make it work for a demo usage, to see if some of you are interested by that kind of things...

If someone wants to give it a try to add some lines of dialogs/a tease to this webpage, you can log in on my IBM Cloud account https://console.bluemix.net/dashboard/apps :
- [email protected]
- YesMistress69

It's a free account without personal infos so you can do whatever you want with it, just try to not break to much stuff... It's actually really easy to rebuild, but that's just not cool. Then you can go to this page:



https://assistant-us-south.watsonplatfo ... ild/dialog

And see how it works and give it a go ;) It's pretty easy to use and there are actually a lot of cool features for us to use (context etc).

For the moment, the things I added to the basic Watson stuff are:
- A timer. You can use it by setting up a text response with these parameters:
::stroke::<milliseconds between strokes/strokes per second>::<x ps/ms>::<duration in s>
For example:
::stroke::ms::300::10 for a stroke each 300ms for 10 seconds
::stroke::ps::5::10 for 5 strokes per second for 10 seconds


::overload::<milliseconds between strokes/strokes per second>::<x ps/m at the start>::<x ps/m at the end>::<duration in s>
For example:
::overload::ms::1000::200:10 for a overload from 1 stroke per second at the beginning and 10 per second at the end of the period of 10 seconds
::overload::ps::1::10:10 for a overload from 1 stroke per second at the beginning and 10 per second at the end of the period of 10 seconds

::pause::<x s> for a pause of x seconds with a timer

- Audio player. You can use it by setting up a text response with these parameters:
::audio::url::<your audio url here>

I can easily add up command if you guys are interested by any other stuff. I might also develop a true tease viewer if enough of you are interested by the idea and want to develop these kind of teases ;)

Anyway, thanks for any feedbacks you might have!!

That's awesome!

Wow

nice!!

who is the girl used as mistress in the demo? I'm keen to try adding a module to the tease, but I wouldn't want to change girls straight up, that would ruin the flow

Razorsedge wrote: Sat Nov 24, 2018 10:37 pm who is the girl used as mistress in the demo? I'm keen to try adding a module to the tease, but I wouldn't want to change girls straight up, that would ruin the flow

Ashley Doll

http://galleries.thisyearsmodel.com/ash ... id=2580075

thanks for that, I've had a good morning playing with the program, but my section still has some issues, so if anyone knows what they are doing, you are welcome to iron out the kinks, it would be great to see what I was missing.

over all this is a very cool concept though, but did anyone else have any issues getting a headset mic to register? not sure what thats all about.

Although I like the idea behind your concept I am no fan of cloud based applications. One day IBM will decide to stop this service or make you pay and everything will be lost

Thank you for your comments!

Thanks Razoredge for giving it a try! I think I solved most of your problems. Most of them were:

-If you put multiple responses under the same response, the tease will just select one randomly/sequantially each time you get to this node, not say one after the other

- Command for strokes are:
::stroke::ps::5::30
Not:
::stroke::5::30
(ps stands for per seconds, ms for milliseconds)

- When doing option choices, you set the response of the user, he can always say one of them instead of clicking, you should check for a response like "I'm edging", instead of a number

- Jumps are fucking nasty with this Assistant stuff... You have to get the hang of it, it's not easy, just look at how I corrected it

- In fact, user don't have a way to interrupt a timer/a suite of responses from the dom. The only way to check how a user is doing so far is by specifically asking him. You don't have to propose options though, you can just do child nodes and check for the user response.

- I don't see why the Speak to text wouldn't work, it works for me on Windows/Firefox/Chrome. Witch platform are you using?

- Thanks a lot for the feedback and the try! Now we have a new tease section ! =D


VoIDoo, you sure got a point on cloud services! And if it's not that, it's just the usage that will go nuts! However I'm pretty sure speech to text and chatbots will just go more and more common and easy to build and cheaper as time go, so we should have an easy backup. But yeah, got your points, teases will be lost :/

Anyway, I think if I continue working on that I will go for a less modulable but more easy to manipulate solution, without using Assistant Dialog (which is nasty to code, not really built for long dialog, more for quick fire questions/answers and not really reliable (nodes keep disappearing/bugging)), using Google Cloud Speech to Text instead of IBM's (it's so much better!) and basically using assistants (from google of IBM or Amazon) just for intentions and entities extractions, and do the tease himself by coding...I'm sure I can do something really awesome there...

Thank you all for your comments, feel free to continue to test that, add teases, and enjoy! ;)

Thanks illusion. I had no idea what I was doing, but enjoyed trying to learn as I went. So your feedback will be really helpful. I really appreciate you going through it. Even as I just read your points then I said “ohh of coarse” to half of them. So I look forward to going through it properly tomorrow.

i cant seem to hear her! :p

Can't speak to her or hear her in any voice, that's sad because that's look amazing

timoth31 wrote: Tue Nov 27, 2018 9:40 am Can't speak to her or hear her in any voice, that's sad because that's look amazing

Tried Chrome first. Had the same problem. Then i tried firefox and it worked.

G3General wrote: Tue Nov 27, 2018 1:41 pm

timoth31 wrote: Tue Nov 27, 2018 9:40 am Can't speak to her or hear her in any voice, that's sad because that's look amazing

Tried Chrome first. Had the same problem. Then i tried firefox and it worked.

ok thanks will try Firefox then :)