Re: [Kaldi-developers] A Question about Kaldi

[Daniel P. <dp...@gm...>]

Sangho,

Kaldi is not limited to use GridEngine, but it will work best with
distributed-computing systems that work on a similar principle to
GridEngine, i.e. that have a concept of jobs and assume a shared file
system.

We did not really consider using Hadoop because it's mainly geared
towards Java, and Java is too slow to do speech recognition.  If there
were a widely used open source implementation of MapReduce that worked
well with C++, we would definitely have considered using it.

Dan


> Dear Kaldi team,
>
>
> Hello, I am a graduate student studying in Sogang University, South Korea.
>
> I and my friends have researched about "DNN-based Acoustic Model Training".
>
> We wonder why you used a Grid Engine instead of using other distributed
> computing tools(eg. Hadoop) for training.
>
> We tried to find it but we couldn't.
>
> The "Kaldi forums on Sourceforge" is not available so we contact inevitably.
>
> Would you mind telling me a reason why you used GE?
>
> Or documents about this are also good for us.
>
> We would appreciate it if you would send your reply.
>
>
>
> Best Regards,
>
> SangHo Wang
>
> ---------------------------------------------------
> DCCLAB Sogang University
> san...@gm...
> +82 (0)10 9808 0510
> ---------------------------------------------------
>
>
> ------------------------------------------------------------------------------
> Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
> from Actuate! Instantly Supercharge Your Business Reports and Dashboards
> with Interactivity, Sharing, Native Excel Exports, App Integration & more
> Get technology previously reserved for billion-dollar corporations, FREE
> http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk
> _______________________________________________
> Kaldi-developers mailing list
> Kal...@li...
> https://lists.sourceforge.net/lists/listinfo/kaldi-developers
>