为你的应用插上语音的翅膀(iOS讯飞语音识别)

一.集成

• 首先到 http://www.xfyun.cn 注册并登陆下

  1 创建新应用(获得后续的appid以及开通服务)

new.jpeg

2 得到appid,点击开通服务选择->语音听写,自动跳转下载sdk界面

chuangjian.jpeg

• 新建工程开始集成

  1.添加系统库<libz.tbd,AVFoundation.framework,SystemConfiguration.framework,Foundation.framework,CoreTelephoney.framework,AudioToolbox.framework,UIKit.framework,CoreLocation.framework,AddressBook.framework,AddressBook.framework,AddressBook.framework>

systemFrame.png

2.添加iflyMSC.framework(在下载的ZIP->lib 目录下)

iflyframe.png

***

二.初始化

1.AppDelegate添加如下代码

    //设置sdk的log等级，log保存在下面设置的工作路径中
        [IFlySetting setLogFile:LVL_ALL];
        
    //    打开输出在console的log开关
        [IFlySetting showLogcat:YES];
    
    //设置sdk的工作路径
    NSArray *paths = NSSearchPathForDirectoriesInDomains(NSCachesDirectory, NSUserDomainMask, YES);
    NSString *cachePath = [paths objectAtIndex:0];
    [IFlySetting setLogFilePath:cachePath];
    
    
    NSString *initString = [[NSString alloc] initWithFormat:@"appid=%@",@"573b1d27"];
    
    
    [IFlySpeechUtility createUtility:initString];




2.讯飞的语音识别主要有2个类 **IFlySpeechRecognizer(不带界面)** 和 **IFlyRecognizerView(界面)**,用法差不多,主要介绍带界面的用法.

- (IFlyRecognizerView *)iflyRecognizerView{
    
    if (!_iflyRecognizerView) {
    
    IFlyRecognizerView *iflyRecognizerView = [[IFlyRecognizerView alloc] initWithCenter:self.view.center];
    iflyRecognizerView.delegate = self;
    [iflyRecognizerView setParameter:@"" forKey:[IFlySpeechConstant PARAMS]];
    
    //设置听写模式
    [iflyRecognizerView setParameter:@"iat" forKey:[IFlySpeechConstant IFLY_DOMAIN]];
    
    
    IATConfig *instance = [IATConfig sharedInstance];
    //设置最长录音时间
    [iflyRecognizerView setParameter:instance.speechTimeout forKey:[IFlySpeechConstant SPEECH_TIMEOUT]];
    //设置后端点
    [iflyRecognizerView setParameter:instance.vadEos forKey:[IFlySpeechConstant VAD_EOS]];
    //设置前端点
    [iflyRecognizerView setParameter:instance.vadBos forKey:[IFlySpeechConstant VAD_BOS]];
    //网络等待时间
    [iflyRecognizerView setParameter:@"20000" forKey:[IFlySpeechConstant NET_TIMEOUT]];
    
    //设置采样率，推荐使用16K
    [iflyRecognizerView setParameter:instance.sampleRate forKey:[IFlySpeechConstant SAMPLE_RATE]];
    if ([instance.language isEqualToString:[IATConfig chinese]]) {
        //设置语言
        [iflyRecognizerView setParameter:instance.language forKey:[IFlySpeechConstant LANGUAGE]];
        //设置方言
        [iflyRecognizerView setParameter:instance.accent forKey:[IFlySpeechConstant ACCENT]];
    }else if ([instance.language isEqualToString:[IATConfig english]]) {
        //设置语言
        [iflyRecognizerView setParameter:instance.language forKey:[IFlySpeechConstant LANGUAGE]];
    }
    //设置是否返回标点符号
    [iflyRecognizerView setParameter:instance.dot forKey:[IFlySpeechConstant ASR_PTT]];
    
    self.iflyRecognizerView = iflyRecognizerView;
    
    
}

return _iflyRecognizerView;


}
````

然后启动服务 [self.iflyRecognizerView start];

3 然后通过代理 IFlyRecognizerViewDelegate 该代理就2个方法如下

/*!
 *  回调返回识别结果
 *
 *  @param resultArray 识别结果，NSArray的第一个元素为NSDictionary，NSDictionary的key为识别结果，sc为识别结果的置信度
 *  @param isLast      -[out] 是否最后一个结果
 */
- (void)onResult:(NSArray *)resultArray isLast:(BOOL) isLast;

/*!
 *  识别结束回调
 *
 *  @param error 识别结束错误码
 */
- (void)onError: (IFlySpeechError *) error;

三.代理模式到Block的转换(简单封装)

1.抽取一个工具类,通过block传递结果
/**
 *  回调
 */
typedef void (^RecognizerBlock)(IFlySpeechError *error,NSString *result);



@interface ZZiflyTool : NSObject

+ (instancetype)shareTool;


- (void)startRecognizer:(RecognizerBlock)block;

@end




外界只需要一句话就可以调用了

[[ZZiflyTool shareTool] startRecognizer:^(IFlySpeechError *error, NSString *result) {
    
    
    if (error) {
        
        [self showMes:error.errorDesc];
        
        return;
    }
    
   NSString *str =  result.length ? result:@"没听到你说啥";
    
    [self showMes:str];
    
}];

DEMO下载: https://github.com/zzBelieve/ifly