Menu

low recognition accuray for the german model

Help
2018-09-21
2018-09-24
  • Mohamed Emara

    Mohamed Emara - 2018-09-21

    Hello,
    I am trying to recognize some German speech using the following language model "cmusphinx-ptm-generic-de-r20180609" but the accuracy is about 35% which is too low.
    I am recording a .wav file by a native german speaker using the following command :
    rec -V1 -q -r 16000 -c 1 -b 16 -e signed-integer --endian little "test1.wav"
    and then run the following command

    pocketsphinx_batch -adcin yes -cepdir wav -cepext .wav -ctl test.fileids -lm "file.lm" -dict "file.dic" -hmm "acoustic_model_directory" -hyp test.hyp

    word_align.pl test.transcription test.hyp

    Is there something wrong becauause I think it's not normal to get such low accuracy using the provided models?

     

Log in to post a comment.

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.