C# return speech phonetic without recognize it

问题

I'm trying to make a human like dictionary (db) of Persian speech, so I tried to speak and have c# produce it phonetically, but the problem is I cant find any event to find what I spoke. For example, we have SpeechRecognized event but it fired after recognizing the speech. Here is my code sample:

<pre>
<code>
SpeechRecognizer rec = new SpeechRecognizer();

public Form1(){

        InitializeComponent();

        rec.SpeechRecognized += rec_SpeechRecognized(rec_SpeechRecognized); // this will fire after recognize
        rec.SpeechDetected +=new EventHandler(rec_SpeechDetected); // this will fired each time but with no return
        rec.enabled = true;
    }
</code>
<pre>

Note: I want c# to produce the phonetic value of what I say, not to recognize it.

回答1:

I don't think System.Speech.Recognition will expose a phonetic interpretation of what you said. The Windows recognizer uses a language specific model to try to match words in the specified language.

The speech engine in Windows 7 supports the following languages: Chinese (Simplified), Chinese (Traditional), French, German, Japanese, Spanish, UK English, and US English. See http://msdn.microsoft.com/en-us/goglobal/ee426904

The Microsoft server speech engine supports 26 languages. I don't believe Persian is supported. See http://www.microsoft.com/downloads/en/details.aspx?FamilyID=F704CD64-1DBF-47A7-BA49-27C5843A12D5

Perhaps using C++ and SAPI you can get to the underlying phonemes. If you search "SAPI Phoneme Extraction" you may find something helpful. In particular look at Speech Recognition with SAPI: Custom Language Support through phenomes which describes building a custom grammar to try to extract phonemes from an alternate language.

Other interesting references I saw at http://developer.valvesoftware.com/wiki/Phoneme_Tool and http://www.mail-archive.com/hlcoders@list.valvesoftware.com/msg19793.html

来源：https://stackoverflow.com/questions/6063823/c-sharp-return-speech-phonetic-without-recognize-it

标签

speech-recognition