About Chinese Pinyin model

Deep Speech
#1

Is there someone experienced using pinyin dataset to train a model, like THCHS-30(https://www.openslr.org/18/), I have a problem with the alphabet set and language model.
I have tried with the officially released alphabet set, and process the format like below picture

42
Screenshot 2019-09-21 at 11.51.42.png446×1044 9.16 KB

but it does not work. Always gives the feedback like
52
Screenshot 2019-09-21 at 11.55.52.png2582×98 14.1 KB
. The transcript of the A22_50.wav is like /DeepSpeech/cn_corpus/THCHS_30/data_thchs30/train/ **A22_50.wav** ,144044,ii i1 x iang1 g ang3 d e5 k ang4 r iz4 j iu4 uu uang2 vv vn4 d ong4 ii i3 j iu4 uu uang2 uu un2 h ua4 vv vn4 d ong4 uu ui2 zh u3 ii iao4 x ing2 sh ix4 and it does not contain the single ‘i’.
Does anybody encounter problem like this?

#2

Hi,

Is your problem solved? My assumption is because the file itself. Which operation system do you use? if it is linux, I think you will have to change the line endings to ‘Unix’ (you can use sublime text to do that easily).

BTW, if your problem is solved. How’s the performance? I am preparing to train one with pinyin as input as well.