[archit507]

Hi Thomas,

Can we do speech recognization like can we convert speech to text using visual prolog and store it in database? If yes, then can you please provide me any document regarding that.

[Thomas Linder Puls]

Yes, that is possible. But not trivial.

You will have to import the Microsoft SAPI COM component in your project.

Notice that working with COM requires the Commercial Edition.

[archit507]

Thank you for the update. Can you please provide me code regarding speech recognization, so that I can refer and work on that.

[Thomas Linder Puls]

This recognizes speech from a file:

Code: Select all

implement main
    open core, sapiDDI_types
 
constants
    recognizerId : string = @"HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Recognizers\Tokens\MS-1033-80-DESK".
    grammarId = 123.
    spfei_interest : unsigned64 = spfei_flagcheck + 2 ^ spei_end_sr_stream + 2 ^ spei_recognition.
 
clauses
    run() :-
        stdio::outputStream:allowNullPointers := true,
        AudioFileName = @"..\test.wav",
        Recognizer = spInprocRecognizer::new(),
        SpToken = spObjectToken::new(),
        RecognizerToken = SpToken:get_ISpObjectToken_import(),
        RecognizerToken:setId(uncheckedConvert(string, null), recognizerId, b_false),
        Recognizer:setRecognizer(RecognizerToken),
        Recognizer:createRecoContext(RecoContext),
        RecoContext:createGrammar(grammarId, Grammar),
        SpStream = spStream::new(),
        SpStream:bindToFile(AudioFileName, spfm_open_readonly, uncheckedConvert(core::nativeGuid, null), uncheckedConvert(waveformatex, null), 0),
        Recognizer:setInput(SpStream:getIUnknown(), 0),
        Grammar:loadDictation("", splo_static),
        Grammar:setDictationState(sprs_active),
        RecoContext:setNotifyWin32Event(),
        RecoContext:setInterest(spfei_interest, spfei_interest),
        try
            eventLoop(RecoContext)
        finally
            Grammar:unloadDictation(),
            Grammar:setDictationState(sprs_inactive),
            Grammar:release(),
            SpStream:release(),
            SpStream:close(),
            RecoContext:release(),
            Recognizer:release()
        end try.
 
class predicates
    eventLoop : (iSpRecoContext_import RecoContext).
clauses
    eventLoop(RecoContext) :-
        std::repeat(), %+
            RecoContext:waitForNotifyEvent(0xFFFFFFFF),
            RecoContext:getEvents(1, Event, Fetched),
            if Fetched <> 0 then
                handleEvent(Event)
            end if,
            (Fetched = 0 orelse spevent(spei_end_sr_stream, _, _, _, _, _) = Event),
        !.
 
    eventLoop(_RecoContext).
 
class predicates
    handleEvent : (sapiDDI_types::spevent Event).
clauses
    handleEvent(sapiDDI_types::spevent(spei_recognition, _ParamType, _StreamNum, _AudioStreamOffset, _WParam, LParam)) :-
        !,
        Native = uncheckedConvert(iSpRecoResult_native, LParam),
        RecoResult = iSpRecoResult_import::new(Native, comMemory::release()),
        RecoResult:getText(0xFFFFFFFF, 0xFFFFFFFF, 1, Text, DisplayAttributes),
        RecoResult:getPhrase(Phrase),
        stdio::writef("Text = %\nDisplayAttributes = %\nPhrase = %\n", toString(Text), DisplayAttributes, Phrase).
 
    handleEvent(_).
 
end implement main

[archit507]

Thank you Thomas.