A local business just offered me $10K to develop a robocaller for them that will just do more than merely play...

A local business just offered me $10K to develop a robocaller for them that will just do more than merely play pre-recorded messages while appearing to be a real human.

I haven't accepted the project yet but I said I'll run some tests and see if the system as a whole works. I've set it a virtual extension in Asterisk to call and talk to the bot.

For Speech to Text, I've found DeepSpeech to be painfully slow even with two GTX 1080 Ti. Google Cloud Speech to Text performed okay but wasn't fast enough either. I've found the Watson API to be really fast.

What is a good API to compare the recognized text against what is in the AIML template? Am I just supposed to have a list of variations in the pattern or is there a more robust way of matching against the template?

Attached: robocaller.png (1340x984, 158K)

Other urls found in this thread:

audacityteam.org/download/linux/
twitter.com/NSFWRedditVideo

Give me $5K and I'll tell you

Use Google Assistant lol

Google Assistant doesn't quite do what the client needs. They need the bot to pitch their product. Depending on what the prospect might say, it will pitch from one of 5-10 variations.

Is there anything like Audacity but for Linux?

Hrm let me think.
Audacity you fuckboi

Throw 5k in my direction and i got u covered

Attached: 1539152175327.jpg (472x461, 30K)

AIML is for static bots. You need thousands of lines of AIML templates to make a useful bot.

You're better off using Tensorflow and NLTK.

Make it 50. This is something worth much more.

this is something you base your entire startup on, not something you do for $10k

Yea, but on Linux. I guess I could always try running Audacity in WINE.

It's meant to work only in a narrow situation and doesn't require a full blown AI. It is mostly gonna play pre-recorded messages. It is definitely doable in a week or two.

Look up Sphinx grammars, it can transcribe based on a series of expected spoken English words. You parse that and then play a pre-recorded text.

Audacity is available on Linux without using wine

I don't want to use a VM just for Audacity though. Is there a Linux alternative?

If it's only going to play pre recorded messages why do you give a shit about DeepSpeech being slow?

yeah here
audacityteam.org/download/linux/

Are you Homer Simpson?

You're actually going to code a robocaller?
have you no ethics?

Attached: 277A571E3EEA46BCAB3991FAA9ACDC65.png (645x729, 107K)

You fucking knob head.

Attached: 1545688868270.jpg (320x301, 22K)

You can run Audacity natively on Linus, but only on Gentoo. So if you install Gentoo you can run Audacity.

Okay, develop it but don't sell the program. You want to get them hooked, so you sell it as a service. $200 per month, you'll pass 10k in less than 5 years. Sell your service to many people, and you'll have your own business.

Ethics are for the poor.
Also this is just setting the stage for the phase two of his master plan, a spoken captcha for phonecalls

>falling for such lowq bait