这是C的代码,供大神们参考。。。
语音识别编程涉及IspRecognizer,IspRecoContext和ISpRecoGrammar等多个语音识别引擎接口。下面先来设计一个操作语音识别的类CSpeechRecognition,然后基于该类来实现一个实例。
11.3.1
构造CSpeechRecognition类
CSpeechRecognition类封装了语音识别操作所需调用的几个接口,使用它进行语音识别编程很方便,也很简洁。
CSpeechRecognition类的定义如下:
///////////////////////////////////////////////////////////////
// active speech engine
#include <atlbase.h>
extern CComModule _Module;
#include <atlcom.h>
#include <sapi.h>
#include <sphelper.h>
#include <spuihelp.h>
///////////////////////////////////////////////////////////////
// speech message
#define WM_SREVENT
WM_USER+102
class CSpeechRecognition
{
public:
CSpeechRecognition();
virtual ~CSpeechRecognition();
// initialize
BOOL Initialize(HWND hWnd = NULL, BOOL bIsShared = TRUE);
void Destroy();
// start and stop
BOOL Start();
BOOL Stop();
BOOL IsDictationOn()
{
return m_bOnDictation;
}
// event handler
void GetText(WCHAR **ppszCoMemText, ULONG ulStart = 0, ULONG nlCount = -1);
// voice training
HRESULT VoiceTraining(HWND hWndParent);
// microphone setup
HRESULT MicrophoneSetup(HWND hWndParent);
// token list
HRESULT InitTokenList(HWND hWnd, BOOL bIsComboBox = FALSE);
// error string
CString GetErrorString()
{
return m_sError;
}
// interface
CComPtr<ISpRecognizer> m_cpRecoEngine;
// SR engine
CComPtr<ISpRecoContext> m_cpRecoCtxt;
//Recognition contextfor dictation
CComPtr<ISpRecoGrammar> m_cpDictationGrammar;
// Dictation grammar
private:
CString m_sError;
BOOL
m_bOnDictation;
};
其中定义的消息WM_SREVENT用于指示语音识别事件,该消息将通知到初始化函数指定的响应窗口。
类中定义了3个接口指针m_cpRecoEngine,m_cpRecoCtxt和m_cpDictationGrammar,分别用于引用语音识别引擎的3个重要接口IspRecognizer,ISpRecoContext和IspRecoGrammar。
初始化函数Initialize设定了语音识别引擎的基本工作环境,包括引擎、识别上下文、语法、音频和事件等的初始化:
BOOL CSpeechRecognition::Initialize(HWND hWnd, BOOL bIsShared)
{
// com library
if (FAILED(CoInitialize(NULL)))
{
m_sError=_T("Error intialization COM");
return FALSE;
}
// SR engine
HRESULT hr = S_OK;
if (bIsShared)
{
// Shared reco engine.
// For a shared reco engine, the audio gets setup automatically
hr = m_cpRecoEngine.CoCreateInstance( CLSID_SpSharedRecognizer );
}
else
{
hr = m_cpRecoEngine.CoCreateInstance(CLSID_SpInprocRecognizer);
}
// RecoContext
if( SUCCEEDED( hr ) )
{
hr = m_cpRecoEngine->CreateRecoContext( &m_cpRecoCtxt );
}
// Set recognition notification for dictation
if (SUCCEEDED(hr))
{
hr = m_cpRecoCtxt->SetNotifyWindowMessage( hWnd, WM_SREVENT, 0, 0 );
}
if (SUCCEEDED(hr))
{
// when the engine has recognized something
const ULONGLONG ullInterest = SPFEI(SPEI_RECOGNITION);
hr = m_cpRecoCtxt->SetInterest(ullInterest, ullInterest);
}
// create default audio object
CComPtr<ISpAudio> cpAudio;
hr = SpCreateDefaultObjectFromCategoryId(SPCAT_AUDIOIN, &cpAudio);
// set the input for the engine
hr = m_cpRecoEngine->SetInput(cpAudio, TRUE);
hr = m_cpRecoEngine->SetRecoState( SPRST_ACTIVE );
// grammar
if (SUCCEEDED(hr))
{
// Specifies that the grammar we want is a dictation grammar.
// Initializes the grammar (m_cpDictationGrammar)
hr = m_cpRecoCtxt->CreateGrammar( 0, &m_cpDictationGrammar );
}
if
(SUCCEEDED(hr))
{hr = m_cpDictationGrammar->LoadDictation(NULL, SPLO_STATIC);
}
if (SUCCEEDED(hr))
{
hr = m_cpDictationGrammar->SetDictationState( SPRS_ACTIVE );
}
if (FAILED(hr))
{
m_cpDictationGrammar.Release();
}
return (hr == S_OK);
}
释放函数Destroy被类的析构函数调用,释放了类所引用的所有接口:
void CSpeechRecognition::Destroy()
{
if (m_cpDictationGrammar)
m_cpDictationGrammar.Release();
if (m_cpRecoCtxt)
m_cpRecoCtxt.Release();
if (m_cpRecoEngine)
m_cpRecoEngine.Release();
CoUninitialize();
}
函数Start和Stop用来控制开始和停止接受及识别语音,它们通过调用引擎接口的SetRecoState方法来实现:
BOOL CSpeechRecognition::Start()
{
if (m_bOnDictation)
return TRUE;
HRESULT hr = m_cpRecoEngine->SetRecoState( SPRST_ACTIVE );
if (FAILED(hr))
return FALSE;
m_bOnDictation = TRUE;
return TRUE;
}
BOOL CSpeechRecognition::Stop()
{
if (! m_bOnDictation)
return TRUE;
HRESULT hr = m_cpRecoEngine->SetRecoState( SPRST_INACTIVE );
if (FAILED(hr))
return FALSE;
m_bOnDictation = FALSE;
return TRUE;
}
函数GetText是获取从语音中已识别出的文字的关键,应该在响应识别事件/消息的响应函数中调用,其代码如下所示。
void CSpeechRecognition::GetText(WCHAR **ppszCoMemText, ULONG ulStart, ULONG nlCount)
{
USES_CONVERSION;
CSpEvent event;
// Process all of the recognition events
while (event.GetFrom(m_cpRecoCtxt) == S_OK)
{
switch (event.eEventId)
{
case SPEI_RECOGNITION:
// There may be multiple recognition results, so get all of them
{
HRESULT hr = S_OK;
if (nlCount == -1)
event.RecoResult()->GetText(SP_GETWHOLEPHRASE,
SP_GETWHOLEPHRASE, TRUE, ppszCoMemText, NULL);
else
{
ASSERT(nlCount > 0);
event.RecoResult()->GetText(ulStart, nlCount, FALSE,
ppszCoMemText, NULL);
}
}
break;
}
}
}
函数InitTokenList调用SpInitTokenComboBox和SpInitTokenListBox函数来实现语音语言在列表或组合列表中的列表显示和选择:
HRESULT CSpeechRecognition::InitTokenList(HWND hWnd, BOOL bIsComboBox)
{
if (bIsComboBox)
return SpInitTokenComboBox(hWnd, SPCAT_RECOGNIZERS);
else
return SpInitTokenListBox(hWnd, SPCAT_RECOGNIZERS);
}
语音识别涉及语音的输入,通常用话筒来输入语音。进行语音识别前,需要判断话筒的位置和设置是否合理,以保证语音识别引擎能获得有效的语音输入。函数MicrophoneSetup调用语音识别引擎接口的DisplayUI方法来显示一个设置话筒的向导,如图11-4所示。示例代码如下所示:
HRESULT CSpeechRecognition::MicrophoneSetup(HWND hWndParent)
{
return m_cpRecoEngine->DisplayUI(hWndParent, NULL, SPDUI_MicTraining, NULL, 0);
}
语音训练是语音识别的重要基础,为了获得期望的识别效果,必须进行语音训练,以让语音识别引擎熟悉说话者的口音。函数VoiceTraining调用语音识别引擎接口的DisplayUI方法来显示一个语音训练向导,如图11-5所示。示例代码如下所示:
HRESULT CSpeechRecognition::VoiceTraining(HWND hWndParent)
{
return m_cpRecoEngine->DisplayUI(hWndParent, NULL, SPDUI_UserTraining, NULL, 0);
}
与CText2Speech类似,CSpeechRecognition类也提供错误处理机制,由GetErrorString函数可以获得错误信息。
11.3.2
示例:用CSpeechRecognition类编制听写程序
使用CSpeechRecognition类来编写语音识别程序很简单,下面让我们实现一个听写程序Stenotypist,其界面如图11-6所示。
用VisualC++编制Stenotypist的步骤和要点如下:
1)使用AppWizard生成一个基于对话框的项目Stenotypist;
2)将SpeechRecognition.H,SpeechRecognition.CPP增加到Stenotypist项目中;
3)在资源编辑器中编辑好响应的控件;
4)用ClassWizard为控件在CStenotypistDlg 类中生成相应的成员;
5)修改StenotypistDlg.h文件,为类CStenotypistDlg增加相应的变量和函数;
6)用ClassWizard为CStenotypistDlg 类添加对控件和消息的响应函数。StenotypistDlg.h的代码如下。
#include "SpeechRecognition.h"
////////////////////////////////////////////////////////////////////
// CStenotypistDlg dialog
class CStenotypistDlg : public CDialog
{
// Construction
public:
CStenotypistDlg(CWnd* pParent = NULL);
// standard constructor
// Dialog Data
//{{AFX_DATA(CStenotypistDlg)
enum { IDD = IDD_STENOTYPIST_DIALOG };
CButton
m_btDictation;
CString
m_strText;
//}}AFX_DATA
// ClassWizard generated virtual function overrides
//{{AFX_VIRTUAL(CStenotypistDlg)
protected:
virtual void DoDataExchange(CDataExchange* pDX);
// DDX/DDV support
//}}AFX_VIRTUAL
CSpeechRecognition
m_SpeechRecognition;
// Implementation
protected:
HICON m_hIcon;
// Generated message map functions
//{{AFX_MSG(CStenotypistDlg)
virtual BOOL OnInitDialog();
afx_msg void OnSysCommand(UINT nID, LPARAM lParam);
afx_msg void OnPaint();
afx_msg HCURSOR OnQueryDragIcon();
afx_msg void OnButtonVt();
afx_msg void OnButtonMs();
afx_msg void OnButtonDictate();
//}}AFX_MSG
afx_msg LRESULT OnSREvent(WPARAM, LPARAM);
DECLARE_MESSAGE_MAP()
};
注意,在CStenotypistDlg类中定义了一个CSpeechRecognition类的对象。
在OnInitDialog函数中调用CSpeechRecognition的初始化函数和设置语音语言列表:
BOOL CStenotypistDlg::OnInitDialog()
{
CDialog::OnInitDialog();
// Add "About..." menu item to system menu.
// IDM_ABOUTBOX must be in the system command range.
ASSERT((IDM_ABOUTBOX & 0xFFF0) == IDM_ABOUTBOX);
ASSERT(IDM_ABOUTBOX < 0xF000);
CMenu* pSysMenu = GetSystemMenu(FALSE);
if (pSysMenu != NULL)
{
CString strAboutMenu;
strAboutMenu.LoadString(IDS_ABOUTBOX);
if (!strAboutMenu.IsEmpty())
{
pSysMenu->AppendMenu(MF_SEPARATOR);
pSysMenu->AppendMenu(MF_STRING, IDM_ABOUTBOX, strAboutMenu);
}
}
// Set the icon for this dialog.
The framework does this automatically
//
when the application's main window is not a dialog
SetIcon(m_hIcon, TRUE);
// Set big icon
SetIcon(m_hIcon, FALSE);
// Set small icon
// TODO: Add extra initialization here
if (! m_SpeechRecognition.Initialize(m_hWnd))
AfxMessageBox(m_SpeechRecognition.GetErrorString());
m_SpeechRecognition.InitTokenList(GetDlgItem(IDC_LIST1)->m_hWnd);
m_SpeechRecognition.Stop();
return TRUE;
// return TRUE
unless you set the focus to a control
}
开始听写和停止听写的实现较简单,只需调用CSpeechRecognition类的响应函数就能实现,其代码如下所示。注意,停止和开始是互相切换的。
void CStenotypistDlg::OnButtonDictate()
{
if (m_SpeechRecognition.IsDictationOn())
{
m_SpeechRecognition.Stop();
m_btDictation.SetWindowText("听写(&D)");
SetWindowText("听写者 - 请按<听写>按钮开始听写!");
}
else
{
m_SpeechRecognition.Start();
m_btDictation.SetWindowText("停止(&S)");
SetWindowText("听写者 - 正在记录,请口述...");
}
}
设置话筒和语音训练也通过直接调用CSpeechRecognition类的成员函数来实现:
void CStenotypistDlg::OnButtonVt()
{
m_SpeechRecognition.VoiceTraining(m_hWnd);
}
void CStenotypistDlg::OnButtonMs()
{
m_SpeechRecognition.MicrophoneSetup(m_hWnd);
}
为了响应消息WM_SREVENT,需要添加相应的消息响应函数:
BEGIN_MESSAGE_MAP(CStenotypistDlg, CDialog)
//{{AFX_MSG_MAP(CStenotypistDlg)
ON_WM_SYSCOMMAND()
ON_WM_PAINT()
ON_WM_QUERYDRAGICON()
ON_BN_CLICKED(IDC_BUTTON_VT, OnButtonVt)
ON_BN_CLICKED(IDC_BUTTON_MS, OnButtonMs)
ON_BN_CLICKED(IDC_BUTTON_DICTATE, OnButtonDictate)
//}}AFX_MSG_MAP
ON_MESSAGE(WM_SREVENT, OnSREvent)
END_MESSAGE_MAP()
LRESULT CStenotypistDlg::OnSREvent(WPARAM, LPARAM)
{
WCHAR *pwzText;
m_SpeechRecognition.GetText(&pwzText);
m_strText += CString(pwzText);
UpdateData(FALSE);
return 0L;
}
7)为了调用Speech引擎,应该在Microsoft Visual C++编程环境中设置好相应的include和lib设置:
① 设置include路径
●
通过Project→Settings菜单项打开Project Settings对话框;
●
点击C/C++项;
●
在Category下拉列表中选取Preprocessor;
●
在“Additional include directories”编辑框中输入安装Speech SDK的include的路径,默认的路径是C:\Program Files\Microsoft Speech SDK 5.1\Include。
② 设置lib信息
●
通过Project→Settings菜单项打开Project Settings对话框;
●
选择Link项;
●
在Category下拉列表中选取Input项;
●
在“Additional library path”编辑框中输入安装Speech SDK的lib的路径,默认的路径是C:\Program Files\Microsoft Speech SDK 5.1\ Lib\i386;
●
将“sapi.lib”输入“Object/library modules”所标识的编辑框中。
8)编译连接该项目,就可让听写者开始听写了。
Stenotypist项目的所有源代码都存放在附盘的\Source\Stenotypist目录下。