Hi,
I have installed pocketsphinx on an ARM-based board that has not very much proccessing power. When I try to test the pocketsphinx_continuous program, this happens:
Board: ready ...listening
Me: Hi, how are you?
Board: hi ... on who
Me: how are you?
Borad: on ...(some processing)
Me: what's your name?
Board: who are you
Me:what's your name?
Board: (does some proceesing)
Me: how old are you?
Board: wha'ts your name
Me:how old are you?
Board: (some processing)
Me: what's your name
Me:silent
Borad: what's your name
Me:silent
Board:What's your name
Me:silent
Board: how old
Me silent
Board:How old are you
Me: silent
Board: how old are you
Me:silent
Board:what's your name
So, as I tried to give an example, it seems the program buffers my voice in somewhere and tries to analyse them, but it takes a long time and I said other thing in this time.
I want to disable this buffering and make a real-time answer. How is it possible? (I know my board's proccessing power is weak so I like to disable the buffering, maybe it could be work like a real-time app when program waits for finishing the proccessing of previous words, then tries to "listen" and process new words)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Nickolay, Thanks for the above link. Read it but finding it little difficult to deal with. Is there such tuning done in any project? If so, some reference would really be helful. I'm facing the same issue in pocketsphinx android demo project. Do you suggest some workaround?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I have installed pocketsphinx on an ARM-based board that has not very much proccessing power. When I try to test the pocketsphinx_continuous program, this happens:
Board: ready ...listening
Me: Hi, how are you?
Board: hi ... on who
Me: how are you?
Borad: on ...(some processing)
Me: what's your name?
Board: who are you
Me:what's your name?
Board: (does some proceesing)
Me: how old are you?
Board: wha'ts your name
Me:how old are you?
Board: (some processing)
Me: what's your name
Me:silent
Borad: what's your name
Me:silent
Board:What's your name
Me:silent
Board: how old
Me silent
Board:How old are you
Me: silent
Board: how old are you
Me:silent
Board:what's your name
So, as I tried to give an example, it seems the program buffers my voice in somewhere and tries to analyse them, but it takes a long time and I said other thing in this time.
I want to disable this buffering and make a real-time answer. How is it possible? (I know my board's proccessing power is weak so I like to disable the buffering, maybe it could be work like a real-time app when program waits for finishing the proccessing of previous words, then tries to "listen" and process new words)
https://cmusphinx.github.io/wiki/decodertuning/
Hi Nickolay, Thanks for the above link. Read it but finding it little difficult to deal with. Is there such tuning done in any project? If so, some reference would really be helful. I'm facing the same issue in pocketsphinx android demo project. Do you suggest some workaround?