Hello Lucy not detecting

[varadwattamwar]

I completed all the steps successfully but after speaking Hello Lucy there is no response from MM. Mic and speaker is set properly.

[sdetweil]

have you tested the mic this way

arecord test.wav
speak something
ctrl-c to stop recording
aplay test.wav

do you hear what you said?

[varadwattamwar]

aplay test.wav

Yes, I heard what i said, but there is some noise disturbance also.

[sdetweil]

then it will be hard for computer to sound match.. u need to get rid of the noise

[varadwattamwar]

then it will be hard for computer to sound match.. u need to get rid of the noise

Ok alright, need to try with a new mic then.
Thanks for the reply.

[varadwattamwar]

then it will be hard for computer to sound match.. u need to get rid of the noise

then it will be hard for computer to sound match.. u need to get rid of the noise

Ok alright, need to try with a new mic then.
Thanks for the reply.

Still the result is same , no response from lucy.
One thing I observed is, I also get an error for Hello-Lucy while running MM.
Error: Failed to fetch git data for Hello-Lucy: Error: block timeout reached
Is this the reason for no response?

[sdetweil]

Error: Failed to fetch git data for Hello-Lucy: Error: block timeout reached
Is this the reason for no response?

the timeout on the updatenotifications module is too short.. fixed in the upcoming release, and in the magicmirror develop branch now

when MagicMirror is running, in a terminal window or ssh session window do

ps -ef | grep arecord | grep -v grep 

it should put out a line like this

????     11505 17306  0 05:11 pts/9    00:00:45 arecord -D default -q -r 16000 -c 1 -t wav 

if it does not, then it means arecord failed to start
in my case it is using the default device...
arecord -l (lower case L, for list)
will show the devices... i don't know how Lucy is configured for mic..

[varadwattamwar]

Error: Failed to fetch git data for Hello-Lucy: Error: block timeout reached
Is this the reason for no response?

the timeout on the updatenotifications module is too short.. fixed in the upcoming release, and in the magicmirror develop branch now

when MagicMirror is running, in a terminal window or ssh session window do

ps -ef | grep arecord | grep -v grep 

it should put out a line like this

????     11505 17306  0 05:11 pts/9    00:00:45 arecord -D default -q -r 16000 -c 1 -t wav 

if it does not, then it means arecord failed to start
in my case it is using the default device...
arecord -l (lower case L, for list)
will show the devices... i don't know how Lucy is configured for mic..

There is only one device mic shown in the list whose card and device number is 1,0.
Still lucy is not detecting users voice so any other solution?
My Rpi version 3B+.

[varadwattamwar]

I tried:
In asoundrc file I changed “hw:1,0” to “plughw:1,0”
But the result is same lucy is not responding.
Please can you give another solution @mykle1

[sdetweil]

after changing .asoundrc did u reboot and also do the arecord test.wav/aplay test.wav cycle?

[sdetweil]

i would change it back to how it was.. and reboot..

i do not know.. voice detection is done by an outside module.. pocket_sphinx_continuous
i stopped using Lucy because it didn't recognize my speech 90% of the time.

you could add

debug:true 

to the config:{} section for the Hello-Lucy module in config.js and
then check the window where u start MM, either npm start or pm2 (using the pm2 logs --lines=100 command)
to see what, if anything, was recognized..

if you use npm start, u can redirect the output to a file , with npm start >somefile
and then u can look at somefile live or after stopping MM..

[varadwattamwar]

I identified the issue .
The issue is lucy is not correctly interpreting my voice.
Ex: If I say "sleep" It interprets "eight"
This is the issue
Is there any solution for this ?
Is this the fault of the repository or mic?
But my mic is fine.

i would change it back to how it was.. and reboot..

i do not know.. voice detection is done by an outside module.. pocket_sphinx_continuous
i stopped using Lucy because it didn't recognize my speech 90% of the time.

you could add

debug:true 

to the config:{} section for the Hello-Lucy module in config.js and
then check the window where u start MM, either npm start or pm2 (using the pm2 logs --lines=100 command)
to see what, if anything, was recognized..

if you use npm start, u can redirect the output to a file , with npm start >somefile
and then u can look at somefile live or after stopping MM..

[sdetweil]

according to the cmu sphinx doc, you could try to create a new acoustic model

https://stackoverflow.com/questions/33453691/cmu-sphinx-for-indian-english

which leads to this
https://cmusphinx.github.io/wiki/tutorialadapt/
and here
https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/

lucy module expects the ldc/lm files to be named Hello-Lucy.dlc and Hello-Lucy.lm
but I think it regenerates based on the word.json file for usage specific recognition
using lmtool (http://www.speech.cs.cmu.edu/tools/lmtool-new.html)

I don't know how to do the adaptation

[varadwattamwar]

word

Thankyou will try this.

[cq0q]

word

Thankyou will try this.

Hi sir, i have the same problem now .. which is when i open my magic mirror it shows up: command something then it shows error .. and i can't say anything ..
Did you solve the problem ? If you did please tell me how

[TwitchPlaysSpeedruns]

I have an issue where the screen displays a microphone symbol and the word Error. Lucy speaks on start up but does not respond. I have done the test and can record and playback in cmd. Using Rpi3 and raspian full.

card 1: Seri [Plantronics Blackwire 3220 Seri], device 0: USB Audio [USB Audio]
Subdevices: 1/1
Subdevice #0: subdevice #0

I have set the config to both (1,0) and (0,0) and both have the same error.
"ps -ef | grep arecord | grep -v grep" does not return anything

Any ideas?

[roccamorice]

I have the exact same issue/behaviour as TwitchPlaysSpeedruns. Was this ever solved?

[sdetweil]

@roccamorice did you try arecord test.wav

then aplay test.wav

[roccamorice]

Yes. It worked fine.Sent from my iPhoneOn Feb 16, 2024, at 9:26 PM, sam detweiler ***@***.***> wrote: @roccamorice did you try arecord test.wav then aplay test.wav —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***>

[roccamorice]

Also when I added the Debug: true to the module config, I don't get a debug log although the display that normally shows icon of microphone and ERROR also displays UNDEFINED below the microphone icon. I only see the error.log in the Hello-Lucy modules directory and the error I keep getting is " Error: spawn pocketsphinx_continuous ENOENT -2".

[sdetweil]

@roccamorice did you do step 3 of the installation

[roccamorice]

Yes I did step 3 and the dependencies.sh executed successfully.

[sdetweil]

step 3 should have installed the pocket sphinx continuous command line app.

it may not have updated the path. I would see if you can find the executable...

[roccamorice]

there is a pocketsphinx-continuous directory in the node_modules directory in Hello-Lucy modules directory. In this directory there are index.js and package.json files. I don't think you are referring to these as the executable...or are you?

[sdetweil]

well, it looks like they have removed that program.. SO.. rework is required on Hello-Lucy

https://cmusphinx.github.io/2022/08/pocketsphinx-continuous/