Natural Reader Speech

479 views
Skip to first unread message

Autotel 1

unread,
Jan 3, 2017, 2:15:59 PM1/3/17
to InMoov

Hello all
I am re installing everything on a new PC and I am now struggling with which version and sketch I should be using. Older sketches using Acapela speech now give me the infringement message so i tried using the Mary speech but I could not get that running at all. Looking at these posts it appears you are looking to use Natural reader Speech now  so I set out trying to use this. 

Here  is what I have 

Latest version of MRL 1861

I have tried inmoov3.deep.AB.V7.py and tried to modify to use natural reader speech but this gets as far as speech commands but every command is answered with 
"I have no answer that " I cant get into Gestures and the only command that works is "Hello"

I have also tried  inmoov3.deep.AB.V9.py with similar results to above.

If anybody can give me an idea of which sketch I should be using, and how to get Natural Reader speech and Gestures working that would be great.

Thanks 

Richard

gael langevin

unread,
Jan 3, 2017, 5:02:25 PM1/3/17
to Autotel 1, InMoov
@Autotel, remove all aimlif from :
C:\Myrobotlab\myrobotlab.1.0.1861\ProgramAB\bots\inmoovWebKit\aimlif
And then use this script:
https://github.com/MyRobotLab/pyrobotlab/blob/master/home/hairygael/InMoov3.Deep.AB.V9.py

Remember version 1861 is a beta version and has issues with gui hanging during launching InMoov. Attach and Detach servo is also having issues.
To run the version without running the robot but only the chatbot, modify:
# toggle to only load program ab and skip the inmoov services
startInMoov = False



To use version 1723 (Kraken) with MaryTTS instead of Acapela modify:
#############################
# LANGUAGE ( FR/EN )
#lang="EN" #for NaturalReaderSpeech
lang="en_US" #for MarySpeech
#Voice="Ryan" # Bruno in French for NaturalReaderSpeech
Voice="cmu-slt-hsmm" #for MarySpeech
#Voice="upmc-pierre-hsmm" #French voice
voiceType = Voice


And modify:
#############################
mouth = Runtime.createAndStart("i01.mouth", "MarySpeech")
#mouth.setVoice(botVoice)
mouth.setVoice(voiceType)
#Volume(amount=2.0)
#TractScaler(amount=1.4)
#F0Scale(f0Add=2.0)
#F0Add(f0Add=60.0)
#Robot(amount=0.0)
#Rate(amount=3.75)
#Whisper(amount=100.0)
#Stadium(amount=100.0)
#Chorus(delay=1.466;amp1=0.54)
#FIRFilter(type=3;fc1=500.0;fc2=2)
#mouth.setAudioEffects("Volume(amount=1.70) + TractScaler(amount=0.95) + F0Scale(f0Add=3.0)")
#mouth.setLanguage(lang)


Gael Langevin
Creator of InMoov
InMoov Robot
@inmoov



--
You received this message because you are subscribed to the Google Groups "InMoov" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+unsubscribe@googlegroups.com.
To post to this group, send email to inm...@googlegroups.com.
Visit this group at https://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/d/optout.

Autotel 1

unread,
Jan 4, 2017, 4:24:31 PM1/4/17
to InMoov
Hell Gael
Thank you fort that I know you must be very busy. I still can't get it to run same results. However I am getting a "no gui" error in the audiofile tab of MRL so I must either be missing something or its in the wrong place. For natural Reader Speech should I have lots of audio files in the MP3 folder?

Thanks Richard

Acapulco Rolf

unread,
Jan 4, 2017, 4:48:49 PM1/4/17
to InMoov
Hi Autotel

Did you follow my step by step instructions here

https://groups.google.com/forum/m/#!topic/inmoov/3VKJA-8FGTc

gael langevin

unread,
Jan 5, 2017, 4:26:07 AM1/5/17
to Acapulco Rolf, InMoov
Hello Autotel,
The folder MP3 doesn't store files automatically. It's a place where you can add manually some mp3, a song for exemple, therefore you can get it played by your robot using it in a python gesture.

NaturalReader will store it's own mp3 files in the directory audiofile/naturalreader/ryan
MaryTTS doesn't store mp3 files

The audiofile tab in MRL doesn't display anything besides "no gui", it is not an error, there is just no visual gui for it.

The default MaryTTS voice, which is "cmu-slt-hsmm" is a female voice, which sounds very robotic and not very good.
You can use this script to run automatically MaryTTS:
https://github.com/MyRobotLab/pyrobotlab/blob/master/home/hairygael/InMoov3.Deep.AB.V7.py

Gael Langevin
Creator of InMoov
InMoov Robot
@inmoov



Autotel 1

unread,
Jan 5, 2017, 4:07:21 PM1/5/17
to InMoov
Hello Gael
I have given up for now on V9 and gone back to trying to get V7 working with Mary Speech. on  mrl1723. Deleted everything and unzipped the 64 bit file from the inmoov website again. I have tried to do everything as you describe but the script wont get past line 107 and gives me this error.

 python : Traceback (most recent call last): File "<string>", line 107, in <module> AttributeError: 'org.myrobotlab.service.MarySpeech' object has no attribute 'setAudioEffects' at org.python.core.Py.AttributeError(Py.java:205) at org.python.core.PyObject.noAttributeError(PyObject.java:1013) at org.python.core.PyObject.__getattr__(PyObject.java:1008) at org.python.pycode._pyx3.f$0(<string>:343) at org.python.pycode._pyx3.call_function(<string>) at org.python.core.PyTableCode.call(PyTableCode.java:167) at org.python.core.PyCode.call(PyCode.java:18) at org.python.core.Py.runCode(Py.java:1386) at org.python.core.Py.exec(Py.java:1430) at org.python.util.PythonInterpreter.exec(PythonInterpreter.java:267) at org.myrobotlab.service.Python$PIThread.run(Python.java:161)


any help would be appreciated.

Best Regards Richard


On Tuesday, January 3, 2017 at 7:15:59 PM UTC, Autotel 1 wrote:

gael langevin

unread,
Jan 5, 2017, 9:10:21 PM1/5/17
to Autotel 1, InMoov
Argh,
Yes sorry this line is causing the issue, you need to add # at the beginning of the line:
#mouth.setAudioEffects("Volume(amount=1.75) + TractScaler(amount=0.95) + F0Scale(f0Add=3.0)")

I modified the script on github to avoid others running in the same problem.
This line works with versions above 1860, it allows to modulate the voice.



You will see that MarySpeech has a voice that is not pleasant to listen.
To get a better voice.
If you add:
https://github.com/MyRobotLab/pyrobotlab/blob/master/home/hairygael/voice-cmu-bdl-hsmm-5.2.jar
in your directory C:\myrobotlab\myrobotlab.1.0.1723\libraries\jar
Modify this line in your script:

Voice="cmu-slt-hsmm" # Default female for MarySpeech
to
Voice="cmu-bdl-hsmm" # Default female for MarySpeech

Your robot should then have a male voice much better, still using MaryTTS.

When you modify your script you need to remove all aimlif in your directory and restart MRL:
C:\myrobotlab\myrobotlab.1.0.1723\ProgramAB\bots\inmoovWebkit\aimlif
Other wise the robot keeps saying: "I have no answer for that"


Gael Langevin
Creator of InMoov
InMoov Robot
@inmoov



--

Autotel 1

unread,
Jan 7, 2017, 5:20:07 PM1/7/17
to InMoov, rpau...@gmail.com


Thank you Gael

That did it ! 

regards Richard
Reply all
Reply to author
Forward
0 new messages