I am using Keyphrase+menu(single wake word with commands) and Keywords (multiple wake words) for wake up and branching. I am fairly happy with the performance. But I get lot of false calls from background noise and other far sounds. I would like to limit the wake word only to loud volume from the active user and ignore any other distant sounds. I tried changing vad_treshold, but it is still detecting wake word even with whisper. What is the range for vad_threhold. Is there another way to limit wake words to high volume sounds? If i can limit this lot of false calls could be eliminated. Any suggestion would be greatly appreciated.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks Nickolay. You have been exceptional in answering all questions on forums. I have few more questions on the treshold. What is the default value for the vad_threshold. Does it influance any other parameter in addition to volume. I am not getting any detection beyond 4 irrespective of volume. Volume has some influance from 1-4. I wish there was another way to screen for louder commands.
Can volume only be filtered at the code level. I am planning to start contributing towards this excellent project. I have strong background in image processing and hope to add some value. Is there code level documentation for getting started on Pocketspinx project.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am using Keyphrase+menu(single wake word with commands) and Keywords (multiple wake words) for wake up and branching. I am fairly happy with the performance. But I get lot of false calls from background noise and other far sounds. I would like to limit the wake word only to loud volume from the active user and ignore any other distant sounds. I tried changing vad_treshold, but it is still detecting wake word even with whisper. What is the range for vad_threhold. Is there another way to limit wake words to high volume sounds? If i can limit this lot of false calls could be eliminated. Any suggestion would be greatly appreciated.
1.0 - 10.0
No
Thanks Nickolay. You have been exceptional in answering all questions on forums. I have few more questions on the treshold. What is the default value for the vad_threshold. Does it influance any other parameter in addition to volume. I am not getting any detection beyond 4 irrespective of volume. Volume has some influance from 1-4. I wish there was another way to screen for louder commands.
Can volume only be filtered at the code level. I am planning to start contributing towards this excellent project. I have strong background in image processing and hope to add some value. Is there code level documentation for getting started on Pocketspinx project.
2.0
No