My pi4 has been built from scratch not using picroft, but built on raspberrian, and it’s working fine, but now I want to switch it to precise, pocket sphinx is just a little to annoying. Now at the moment I don’t have huge amounts of time to train a custom wake word, I was wondering if there where any other wake words that have been already provided enough training to be reliable I know the four that are on the device setup, but as my pi4 has a female voice the “hey Ezra” is not appealing, even something simple like “Computer” would be fine till I can devote enough time to the training exercise. Any hey would be appriciated
you can find community models here, you are encouraged to submit more samples and someone will train a model for you
thank you I new there must have been something, but the little grey cells don’t always works so good
OK took everyone’s great suggestions, figured my way through most of it to the point of testing so I have been playing around with the trigger and threshold values and well I didn’t want it triggering all the time but I would like it to trigger when its wake word is given. But I did notice one thing from the logs that I don’t know if it a problem or not:
2021-06-18 23:05:07.174 | INFO | 26209 | mycroft.client.speech.listener:create_wakeup_recognizer:365 | creating stand up word engine
2021-06-18 23:05:07.175 | INFO | 26209 | mycroft.client.speech.hotword_factory:load_module:467 | Loading “wake up” wake word via pocketsphinx
2021-06-18 23:16:29.966 | WARNING | 26209 | mycroft.client.speech.listener:run:88 | Audio contains no data.
2021-06-18 23:16:29.970 | INFO | 26209 | mycroft.client.speech.listener:create_wake_word_recognizer:328 | Creating wake word engine
2021-06-18 23:16:29.971 | INFO | 26209 | mycroft.client.speech.listener:create_wake_word_recognizer:351 | Using hotword entry for athena
2021-06-18 23:16:29.972 | WARNING | 26209 | mycroft.client.speech.listener:create_wake_word_recognizer:357 | Threshold is missing falling back to listeners configuration
2021-06-18 23:16:29.973 | INFO | 26209 | mycroft.client.speech.hotword_factory:load_module:467 | Loading “athena” wake word via precise
Warning: Failed to load parameters from /home/vstubbs/.mycroft/precise/athena.pb.params
2021-06-18 23:16:30.900 | INFO | 26209 | mycroft.client.speech.listener:create_wakeup_recognizer:365 | creating stand up word engine
2021-06-18 23:16:30.900 | INFO | 26209 | mycroft.client.speech.hotword_factory:load_module:467 | Loading “wake up” wake word via pocketsphinx
it’s the Warning: Failed to load parameters from /home/vstubbs/.mycroft/precise/athena.pb.params I took all three files from the archive and placed it into the directory with the .pb and .pb txt:
lrwxrwxrwx 1 vstubbs users 42 Jun 18 19:43 athena → /mnt/volume2/Precise-Community-Data/athena
-rw-rw-r-- 1 vstubbs users 26635 Nov 13 2018 athena.pb
-rw-rw-r-- 1 vstubbs users 169 Nov 13 2018 athena.pb.params
-rw-rw-r-- 1 vstubbs users 478108 Nov 13 2018 athena.pbtxt
lrwxrwxrwx 1 vstubbs users 44 Jun 18 19:43 computer → /mnt/volume2/Precise-Community-Data/computer
-rw-rw-r-- 1 vstubbs users 26635 Aug 15 2019 computer-en.pb
-rw-rw-r-- 1 vstubbs users 169 Aug 15 2019 computer-en.pb.params
-rw-rw-r-- 1 vstubbs users 478245 Aug 15 2019 computer-en.pbtxt
-rw-r–r-- 1 vstubbs users 16615 Jun 18 20:00 framefile-1024.dat
-rw-r–r-- 1 vstubbs users 1081 Jun 18 20:00 framegraph-1024.gnu
lrwxrwxrwx 1 vstubbs users 44 Jun 18 19:44 licenses → /mnt/volume2/Precise-Community-Data/licenses
lrwxrwxrwx 1 vstubbs users 50 Jun 18 19:43 not-wake-words → /mnt/volume2/Precise-Community-Data/not-wake-words
drwxrwxr-x 1 vstubbs users 5652 Jun 22 2018 precise-engine
-rw-r–r-- 1 vstubbs users 57853556 Mar 13 19:16 precise-engine_0.2.0_x86_64.tar.gz
Just to make sure the problem was not the pi4, I tried the configuration on my main linux server where I have mycroft installed where there is lots of memory and working audio systems and thad the same results and just to be complete here is the user mycroft config:
{
“max_allowed_core_version”: 21.2,
“listener”: {
“wake_word”: “athena”
},
“hotwords”: {
“athena”: {
“module”: “precise”,
“phonemes”: “AE TH IY N AH”,
“local_model_file”: “/home/vstubbs/.mycroft/precise/athena.pb”,
“sensitivity”: 0.5,
“trigger_level”: 5
}
},
“tts”: {
“mimic”: {
“path”: “/usr/bin/mimic”
}
},
“SkillInstallerSkill”: {
“path”: “/usr/bin/msm”
}
}
The warning can be ignored.
I keep getting “precise is taking too long to load”.
any idea what is wrong?