Menu

2 languages at the same time didn't work!

Help
rezaee
2017-06-23
2018-04-24
  • rezaee

    rezaee - 2017-06-23

    Hi
    I did create a dictionary with 2 languages(English/Persian) at the one file like this:

    بگو        B E G U
    خزنده        KH A Z A N D E
    قدت        GH A D E T
    چنده        CH A N D E
    قد         GH A D
    من        M A N
    شب        SH A B
    hi        H AA Y
    hello        H E L L O
    how        H O V
    are        AA R
    you        Y U
    what         V AA T
    is         I Z
    your         Y O R
    name        N E Y M
    old        O L D
    where         V E R
    from        F E R AA M
    

    And built language model with this online tool http://www.speech.cs.cmu.edu/tools/lmtool-new.html
    But after training my acoustic model with sphinxtrain, the resault only response for persian!

    Is it possible to make a recognition system that response for 2 languages at the same time?

     
    • Nickolay V. Shmyrev

      You need to provide more information to get help on this problem - data, models, command lines.

       
  • rezaee

    rezaee - 2017-09-24

    Thank you Nikolay, after a long time that I was in other project i came back here to complete this project. I will put other information very soon.

     
  • rezaee

    rezaee - 2018-04-24

    Hi dear Nikolay!
    I attached my etc folder to my post. I tried to create language model with the online tools the tutorial suggested, but it only works for Persian languages and it can not recognize English words at the same time.

     
  • rezaee

    rezaee - 2018-04-24

    Also the command I use to test is:

    pocketsphinx_continuous -inmic yes -hmm /home/m/myrobot2/robot/model_parameters/robot.cd_cont_200 -lm /home/m/myrobot2/robot/etc/robot.lm -dict /home/m/myrobot2/robot/etc/robot.di

    And it works. But only recognizes Persian sentences and words.

     
  • rezaee

    rezaee - 2018-04-24

    This is the etc file(I have problem with editing my posts)

     
  • rezaee

    rezaee - 2018-04-24

    I found that the problem is because of my phoneset file. I used my own phoneset for both English and Persian languages as you can see in my question(first post) but it seems language model only accepts Persian words and don't wotk for English words with this phoneset!

    It seems I must use the standard English phoneset(I mean what the pocketsphinx use for English) for my Persian words as well to have a bi-langual model....But I don't know where can I find a complete set of them and an instruction to know how can I write Persian words with that phoneset?

     
  • rezaee

    rezaee - 2018-04-24

    Thank you for the paper Nikolay! But it seems as I said above, I must use the standard phoneset that pocketsphinx use for English language, but don't know where can I find a full list of them? I mean a complete phoneset file of what pocketsphinx use

     
  • rezaee

    rezaee - 2018-04-24

    For example, when I tried to write the Persian words with English letters, and tried to make a language model, the dictionary is look like this:

    A   AH
    A(2)    EY
    AB  AE B
    ABI AE B IY
    AGHAB   AE AH B
    AHLE    AE L
    AIR EH R
    AM  AE M
    AM(2)   EY EH M
    ANGUR   AE N G AH R
    ANJIR   AE N JH AH R
    APPLE   AE P AH L
    ARE AA R
    ARE(2)  ER
    AROUND  ER AW N D
    AROUND(2)   ER AW N
    BABAT   B AE B AH T
    BACK    B AE K
    BALA    B AA L AH
    BEBAR   B IH B AA R
    BECHARKH    B AH CH AA R K
    BECOME  B IH K AH M
    BEGU    B IH G Y UW
    BENEVIS B IY AH N IY V IY Z
    BIRD    B ER D
    BIYA    B AY Y AA
    BLACK   B L AE K
    BLUE    B L UW
    BORO    B ER OW
    BRIGHT  B R AY T
    BROWN   B R AW N
    CHAND   CH AE N D
    CHANDE  CH AE N D
    CHANDOMI    CH AE N D AH M IY
    CHAPETO CH AE P AH T OW
    CHE CH EY
    CHETORE CH EH T AH R
    CHIE    CH AY
    CLASE   K L EY Z
    COLOR   K AH L ER
    COLOR(2)    K AO L ER
    COME    K AH M
    CREEPY  K R IY P IY
    DADS    D AE D Z
    DARI    D AE R IY
    DARK    D AA R K
    DARYA   D AE R AY AH
    DASTE   D EY S T
    DAY D EY
    DO  D UW
    DORE    D AO R
    DUS D AH Z
    EDUCATED    EH JH AH K EY T IH D
    EDUCATED(2) EH JH Y UW K EY T IH D
    ESME    EH Z M
    ESMET   EH Z M AH T
    ESMETO  EH Z M AH T OW
    FATHERS F AA DH ER Z
    FAVORITE    F EY V ER IH T
    FAVORITE(2) F EY V R AH T
    FIGS    F IH G Z
    FOOD    F UW D
    FORWARD F AO R W ER D
    FROM    F R AH M
    GHADE   G EY D
    GHADET  G AE D AH T
    GHAHVEI G AE V EY
    GHAZAEI G AE Z IY IY
    GHERMEZ G AH R M AH Z
    GO  G OW
    GOLABI  G AA L AH B IY
    GRADE   G R EY D
    GRAPE   G R EY P
    GRAY    G R EY
    GREEN   G R IY N
    HALET   HH AE L AH T
    HAND    HH AE N D
    HAVA    HH AE V AH
    HELLO   HH AH L OW
    HELLO(2)    HH EH L OW
    HI  HH AY
    HOME    HH OW M
    HOW HH AW
    HULU    HH Y UW L UW
    I   AY
    IN  IH N
    IS  IH Z
    JELO    JH EH L OW
    KHAK    K HH AE K
    KHAKESTARI  K HH AH K EH S T AH R IY
    KHAMUSH K HH AE M AH SH
    KHAZANDE    K HH AE Z AH N D
    KHODET  K HH AA D AH T
    KHUNATUN    K HH Y UW N AH T AH N
    KOJAEI  K AA JH IY IY
    KOJAS   K AA JH AH Z
    KON K AA N
    LEFT    L EH F T
    LOVE    L AH V
    MADARET M AE D AH R AH T
    MAMANET M AE M AH N AH T
    MAN M AE N
    MANO    M AA N OW
    ME  M IY
    MESHKI  M EH SH IY
    MOALEMET    M OW L AH M AH T
    MOMS    M AA M Z
    MOTHERS M AH DH ER Z
    NAME    N EY M
    NARENJI N AH R EH N JH IY
    NIGHT   N AY T
    NOGHREI N AA R IY
    OF  AH V
    OFF AO F
    OLD OW L D
    ON  AA N
    ON(2)   AO N
    OR  AO R
    OR(2)   ER
    ORANGE  AO R AH N JH
    ORANGE(2)   AO R IH N JH
    PARANDE P AE R AH N D
    PEACH   P IY CH
    PEAR    P EH R
    PEDARET P EH D AH R AH T
    PINK    P IH NG K
    PORTEGHAL   P AO R T AH AH L
    PUT P UH T
    RANGIE  R EY N JH IY
    RASTETO R AE S T AH T OW
    RED R EH D
    RIGHT   R AY T
    RO  R OW
    ROSHAN  R AA SH AH N
    ROSHANE R AH SH EY N
    RUZ R AH Z
    SABZ    S AE B Z
    SAFHATO S AE F AH T OW
    SALAM   S AA L AA M
    SALETE  S AH L IY T
    SAVAD   S AE V AH D
    SCREEN  S K R IY N
    SEA S IY
    SEFID   S EH F AH D
    SHAB    SH AE B
    SHO SH OW
    SIB S IH B
    SILVER  S IH L V ER
    SIYAH   S AY AY AH
    SOIL    S OY L
    SURATI  S UH R AH T IY
    TALL    T AO L
    TARIK   T AE R AH K
    TEACHERS    T IY CH ER Z
    TELL    T EH L
    THE DH AH
    THE(2)  DH IY
    TURN    T ER N
    TUSI    T Y UW S IY
    UP  AH P
    WATER   W AO T ER
    WHAT    W AH T
    WHAT(2) HH W AH T
    WHERE   W EH R
    WHERE(2)    HH W EH R
    WHITE   W AY T
    WHITE(2)    HH W AY T
    WRITE   R AY T
    YA  Y AA
    YE  Y IY
    YE(2)   Y EH
    YELLOW  Y EH L OW
    YOU Y UW
    YOUR    Y AO R
    YOUR(2) Y UH R
    YOURSELF    Y ER S EH L F
    YOURSELF(2) Y UH R S EH L F
    YOURSELF(3) Y AO R S EH L F
    ZARD    Z AA R D
    

    But it uses incorrect phonemes for Persian words, because it considers them as English words. So the only way is to correcting them by hand, but I have no reference to the phone lists to know how to correct Persian words phones?

     

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.