Tatiana Likhomanenko

The problem could be in kenlm compilation. Could you use exactly the same version we use in the docker and follow installation in the docker https://github.com/facebookresearch/wav2letter/blob/master/Dockerfile-CUDA-Base#L84 ? For your custom...

Just try to run installation like this ``` cd /root && git clone https://github.com/kpu/kenlm.git && \ cd kenlm && git checkout e47088ddfae810a5ee4c8a9923b5f8071bed1ae8 && \ mkdir build && cd build &&...

Are you using s2s model or ctc? Here is docs on decoder parameters (in which cases what you need to tune) https://github.com/facebookresearch/wav2letter/wiki/Beam-Search-Decoder. Seems you are using s2s, then try to...

@GowthamS07, So as I am understanding you want to have not just Viterbi path. Instead you want to have like zero lm decoding where you still restrict to the lexicon?...

Let me give more details. > **Question#1**: What is the difference between Decoder and Test command while keeping every parameter same in the config file? (using a greedy path for...

Online inference is processing data by chunks and history removed from the beam-search decoder, so we don't fix previous output on the fly as we consume more future. If you...

Hi @realbaker1967, Converted arpa file into bin is not a problem, should be fine - here you can try to specify directly arpa file, w2l works with arpa too (just...

@realbaker1967 > Sorry for the late response. I've tried your suggestions, here the observations: Sure, no problem. - To check if your ngram model is word-based (just to make sure...

Currently we are merging code from w2l into flashlight, so doing git pull on flashlight is taking new version (which has another flags of building). You have two options: 1)...

I guess you installed latest master of flashlight where python bindings now live. Then yes, this is https://github.com/facebookresearch/flashlight/blob/master/docs/source/bindings/python.rst the updated doc how to use them now in flashlight. Mostly the...