Sphinx 3 multi core processing

Speech Recognition Toolkit

Brought to you by: air, arthchan2003, awb, bhiksha, and 5 others

This project can now be found here.

Sphinx 3 multi core processing

Forum: Help

Created: 2011-05-02

Updated: 2012-09-22

binac - 2011-05-02

I am using Sphinx 3 for batch processing in parallel multi core environment.
Problem is that every parallel task must initialize new decoder with same
parameters. This leads to duplicate data that allocate near 2 GB of RAM for 4
core processor in my case.
Is it possible for multiple decoders in separate threads to use same memory
for storing language models and dictionary. Which parameters do I need to
share between decoders.

Thanks,
Binac

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Nickolay V. Shmyrev - 2011-05-02

From the theoretical point of view there is no issue to share the acoustic and
the language model data, there is also a possibility to share the scoring
across multiple streams.

There is no option to do that currently with sphinx3. This feature is not
implemented.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

binac - 2011-05-03

Thanks for reply
Can you please explain how to share scoring from multiple streams?

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Log in to post a comment.