Re: [Kaldi-users] Memory requirement for FSTs

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

BTW, one option is to decode with a fairly tightly pruned LM, and then
rescore the lattices with a larger one.  Search for "lmrescore" in
egs/wsj/s5/run.sh
Dan


On Thu, Oct 17, 2013 at 1:03 AM, Arnab Ghoshal <ar...@gm...> wrote:
> On Thu, Oct 17, 2013 at 8:26 AM, E <oth...@ao...> wrote:
>> but in short why size of HCLG.fst >> sum of size of
>> individual *.fsts)? Is there some redundancy involved?
>
> No redundacny. HCLG is *composition* of the individual FSTs, and its
> size is roughly the product of the sizes of the FSTs being composed
> (modulo determinization & minimization). Try pruning your LM if HCLG
> is too big.
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
> _______________________________________________
> Kaldi-users mailing list
> Kal...@li...
> https://lists.sourceforge.net/lists/listinfo/kaldi-users