Error while adding hot_words , using ds-ctcdecoder 0.9.3 via pip

  File "/usr/local/lib/python3.7/dist-packages/ds_ctcdecoder/swigwrapper.py", line 810, in ctc_beam_search_decoder
return _swigwrapper.ctc_beam_search_decoder(probs, alphabet, beam_size, cutoff_prob, cutoff_top_n, ext_scorer, hot_words, num_results)
TypeError: Wrong number or type of arguments for overloaded function 'ctc_beam_search_decoder'.
  Possible C/C++ prototypes are:
ctc_beam_search_decoder(double const *,int,int,Alphabet const &,size_t,double,size_t,std::shared_ptr< Scorer >,std::unordered_map< std::string,float,std::hash< std::string >,std::equal_to< std::string >,std::allocator< std::pair< std::string const,float > > >,size_t)
ctc_beam_search_decoder(double const *,int,int,Alphabet const &,size_t,double,size_t,std::shared_ptr< Scorer >,std::unordered_map< std::string,float,std::hash< std::string >,std::equal_to< std::string >,std::allocator< std::pair< std::string const,float > > >)

hot_words=dict(hello=1.9)

passing it like this

As I stated on Github, this is not a supported usecase. Can you explain why you are trying this ?

Please also have a look at the AddHotWord binding code, this should show you how to do it.

@lyssyx I have made my very own acoustic model where I get my own logits.
I just want to leverage ctcdecoder from deepspeech. Its working properly except this hot words boost feature.

https://github.com/mozilla/DeepSpeech/blob/ad0f7d2ab7f4505d870fb2eaf25373b788d1d938/native_client/ctcdecode/init.py#L169

Here it is documented about the usage , so thats why I tried using it like that. Confused a little.

As i said, this is not a usecase we can support. Please have at look at the AddHotWord python libdeepspeech binding implementation, it should give you what you need.

That being said, its still weird you want to use ctcdecoder module instead of deepspeech.

i guess this is the part handled by swig: https://github.com/mozilla/DeepSpeech/blob/ad0f7d2ab7f4505d870fb2eaf25373b788d1d938/native_client/ctcdecode/swigwrapper.i#L26