As a non-programmer, I am trying to understand the differences between
Sphinx-4 and PocketSphinx. I understand that they are written in different
languages, and that PocketSphinx is optimized for mobile devices. And
according to http://cmusphinx.sourceforge.net/wiki/versions - PocketSphinx is faster
Would there ever be a reason to run Sphinx-4 on a mobile device? Can
PocketSphinx handle a very large vocabulary? Which would work better (accuracy
and speed) for real time continuous speech?
If it matters - my project requires the processing to be done on the device,
not sent to the cloud. But the device does not have to be an actual phone -
just a pocket sized computer.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm a programmer, and I'm writing a desktop application. I'm comfortable
working in C or in Java, and the non-speech-recognition parts of my
application can be made equally well in either language. I will have a medium-
sized vocabulary (100-ish words) with a relatively complicated grammar (I've
got about 100 lines of a JSGF right now). I am most concerned with ignoring
out-of-grammar utterances, as they will be very common in my application.
Could you compare and contrast sphinx-4 and pocketsphinx for me? Is there one
that I should prefer over the other? Why?
Thanks for the help!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
As a non-programmer, I am trying to understand the differences between
Sphinx-4 and PocketSphinx. I understand that they are written in different
languages, and that PocketSphinx is optimized for mobile devices. And
according to http://cmusphinx.sourceforge.net/wiki/versions - PocketSphinx is faster
Would there ever be a reason to run Sphinx-4 on a mobile device? Can
PocketSphinx handle a very large vocabulary? Which would work better (accuracy
and speed) for real time continuous speech?
If it matters - my project requires the processing to be done on the device,
not sent to the cloud. But the device does not have to be an actual phone -
just a pocket sized computer.
With the current state of mobile device performance there is no reason
Pocketsphinx works with large vocabulary models
Pocketsphinx
Thank you, that's very helpful.
How large can the vocabulary models be for PocketSphinx?
Pocketsphinx supports up to 60k words, using more words is not practical
because of lower accuracy
I have a very similar question.
I'm a programmer, and I'm writing a desktop application. I'm comfortable
working in C or in Java, and the non-speech-recognition parts of my
application can be made equally well in either language. I will have a medium-
sized vocabulary (100-ish words) with a relatively complicated grammar (I've
got about 100 lines of a JSGF right now). I am most concerned with ignoring
out-of-grammar utterances, as they will be very common in my application.
Could you compare and contrast sphinx-4 and pocketsphinx for me? Is there one
that I should prefer over the other? Why?
Thanks for the help!
Please see the documentation
http://cmusphinx.sourceforge.net/wiki/versions