I have downloaded the eng2102 model and when I run the following command
python -m allosaurus.run --lang eng --model eng2102 -i 'my_audio.wav' allosaurus outputs the results with the default phones inventory, i.e all the IPA symbols and not those of the english inventory only . For example, I get "d͡ʒ" which is in the IPA inventory but not in the english one.
Originally, I had the problem using the Recognizer and looking at app.py I realised that some options were by default using 'ipa' and that we could not change them if I'm not wrong. Same in factory.py in the read_lm function. And in a comment it is said 'only phone_ipa model is supported for allosaurus now'.
So is it the normal behavior for the moment, or does the problem comes from some default arguments or am I missing something ?
Good afternoon,
I have downloaded the eng2102 model and when I run the following command
python -m allosaurus.run --lang eng --model eng2102 -i 'my_audio.wav'
allosaurus outputs the results with the default phones inventory, i.e all the IPA symbols and not those of the english inventory only . For example, I get "d͡ʒ" which is in the IPA inventory but not in the english one.Originally, I had the problem using the
Recognizer
and looking at app.py I realised that some options were by default using 'ipa' and that we could not change them if I'm not wrong. Same in factory.py in theread_lm
function. And in a comment it is said 'only phone_ipa model is supported for allosaurus now'.So is it the normal behavior for the moment, or does the problem comes from some default arguments or am I missing something ?
Thank you very much !