Mycroft AI Mimic Speech

as of build 1891...

There's a new speech service.. MimicSpeech.  MimicSpeech is a wrapper around the "mimic" speech project from the Mycroft AI project.

At a friends birthday party recently someone said I should look into the Mycroft AI project.   So, I did, it looks like they've done a good job running with much of the CMU speech stuff.  

So,  In that project there's speech synthesis (and speech recognition and a few other things)  but it seems that it's all in C.   They have a subproject called "mimic" which is a wrapper around a CMU speech to text engine .  They have a nice slack channel.  The community was responsive and supportive. .

As it turns out, you can build the "mimic" binary from source and add that to the MRL repo per operating system.  I've done it for windows 64 bit currently.  (built using cygwin, so i don't know how cross platform it is.)

I've worked up a service for it . the new MimicSpeech service.  It has a few voices built in.  They are decent quality.  The MRL service runs the command line tool  "mimic" to produce the speech.

If we think it's worthwhile to keep as a service in MRL, we'll need to build it on all the platforms that we want to support.  Now, if you're running the windows 64 bit, the mimic service should install and the speakBlocking method on it should work..  If it seems worky for folks, i'll finish flushing out the rest of the methods for it.

More info on the mycroft ai site.

feedback welcome!

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
GroG's picture

Cool Stuff Kwatters ! I've

Cool Stuff Kwatters !

I've crossed their site quite a while back, funny, it was because someone from Mycroft was asking questions about MaryTTS and I'm on the MaryTTS forum mail ...  very cool

I've seen pocket-sphinx demos which appear better than Sphinx4 .. which I never understood (perhaps the cell phone industry drove it faster - but why not Java if you can do that on Android ... I dunno, but I'm usually disappointed by Sphinx4...   it looks like thiers new work, or development .. but not so much ...

I'm excited about checking the mimic service out !


GroG's picture

Great ..  sounds very

Great .. 
sounds very robotic, but its clear 

hairygael's picture

Great work Kwatters! As Grog

Great work Kwatters!

As Grog says, they do sound a bit robotic, but maybe it's easier to improve than MaryTTS. That's only a suggestion, I have no idea how this is actually built and done.

I'm adding the link to the various voices available:

It will save others time for to select a voice test.