前言
语音识别现在已经在开发中越来越常见了，科大讯飞、百度等第三方库层出不穷，在这里简单的介绍一下iOS原生的语音识别该怎么做。其原理就是录音，然后把录音数据传给iOS内部的语音识别库，然后再导出来（原生语音识别所支持的版本为10.0及以上）。
实现
添加配置文件
在info.plist的文件中添加语音识别和麦克风使用的配置Privacy - Speech Recognition Usage Description和Privacy - Microphone Usage Description，如下图：
导入类库
#import <Speech/Speech.h>
#import <AVFoundation/AVFoundation.h>
定义相应的控件
@property (strong, nonatomic) SFSpeechRecognizer *speechRecognizer;
@property (strong, nonatomic) SFSpeechRecognitionTask *recognitionTask;
@property (strong, nonatomic) SFSpeechAudioBufferRecognitionRequest *recognitionRequest;
@property (strong, nonatomic) AVAudioEngine *audioEngine;
@property (strong, nonatomic) AVAudioSession *audioSession;
上面的三个是iOS的语音识别的类,下面两个是音频的类，对AVAudioEngine有兴趣的小伙伴可以去看一下说明1和说明2，对AVAudioSession有兴趣的小伙伴可以去看一下这里。
然后懒加载初始化相应的类，如下：
- (SFSpeechRecognizer *)speechRecognizer {
if (!_speechRecognizer) {
NSLocale *local = [[NSLocale alloc]initWithLocaleIdentifier:@"zh_CN"];
_speechRecognizer = [[SFSpeechRecognizer alloc]initWithLocale:local];
_speechRecognizer.delegate = self;
}
return _speechRecognizer;
}
- (AVAudioEngine *)audioEngine {
if (!_audioEngine) {
_audioEngine = [[AVAudioEngine alloc]init];
}
return _audioEngine;
}
- (AVAudioSession *)audioSession {
if (!_audioSession) {
_audioSession = [AVAudioSession sharedInstance];
NSError *error;
[_audioSession setCategory:AVAudioSessionCategoryRecord error:&error];
[_audioSession setMode:AVAudioSessionModeMeasurement error:&error];
[_audioSession setActive:YES withOptions:AVAudioSessionSetActiveOptionNotifyOthersOnDeactivation error:&error];
}
return _audioSession;
}
小编在初始化语音识别类的时候使用的是zh_CN,这个的意思是语音识别的类型是中文，可以根据需要替换成其他类型。可以通过下列代码获得所支持的类型：
for (NSLocale *temp in SFSpeechRecognizer.supportedLocales) {
NSLog(@"国家代码：%@,语言代码：%@,输入方式:%@_%@", temp.countryCode, temp.languageCode, temp.languageCode, temp.countryCode);
}
可以使用的类型为输入方式后面的输出。
获取相应的授权
- (void)viewDidLoad {
[super viewDidLoad];
[self accessPermissions];
}
- (void)accessPermissions {
// 语音识别授权
[SFSpeechRecognizer requestAuthorization:^(SFSpeechRecognizerAuthorizationStatus status) {
BOOL isEnabled = NO;
NSString *str;
switch (status) {
case SFSpeechRecognizerAuthorizationStatusNotDetermined:
isEnabled = NO;
str = @"不支持录音";
NSLog(@"结果未知 用户尚未进行选择");
break;
case SFSpeechRecognizerAuthorizationStatusDenied:
isEnabled = NO;
str = @"不支持录音";
NSLog(@"用户未授权使用语音识别");
break;
case SFSpeechRecognizerAuthorizationStatusRestricted:
isEnabled = NO;
str = @"不支持录音";
NSLog(@"设备不支持语音识别功能");
break;
case SFSpeechRecognizerAuthorizationStatusAuthorized:
isEnabled = YES;
str = @"开始录音";
NSLog(@"用户授权语音识别");
break;
default:
break;
}
dispatch_async(dispatch_get_main_queue(), ^{
self.recordButton.enabled = isEnabled;
if (isEnabled) {
[self.recordButton setTitle:str forState:UIControlStateNormal];
}else {
[self.recordButton setTitle:str forState:UIControlStateDisabled];
}
});
}];
// 麦克风使用授权
if ([self.audioSession respondsToSelector:@selector(requestRecordPermission:)]) {
[self.audioSession performSelector:@selector(requestRecordPermission:) withObject:^(BOOL granted) {
dispatch_async(dispatch_get_main_queue(), ^{
self.recordButton.enabled = granted;
if (granted) {
NSLog(@"麦克风授权");
[self.recordButton setTitle:@"开始录音" forState:UIControlStateNormal];
}else {
NSLog(@"麦克风未授权");
[self.recordButton setTitle:@"麦克风未授权" forState:UIControlStateDisabled];
}
});
}];
}
}
语音识别
开始录音
- (void)startRecording {
if (_recognitionTask) {
[_recognitionTask cancel];
_recognitionTask = nil;
}
_recognitionRequest = [[SFSpeechAudioBufferRecognitionRequest alloc]init];
AVAudioInputNode *inputNode = self.audioEngine.inputNode;
_recognitionRequest.shouldReportPartialResults = YES;
_recognitionTask = [self.speechRecognizer recognitionTaskWithRequest:_recognitionRequest resultHandler:^(SFSpeechRecognitionResult * _Nullable result, NSError * _Nullable error) {
BOOL isFinal = NO;
if (result) {
self.displayLabel.text = result.bestTranscription.formattedString;
isFinal = result.isFinal;
}
if (error || isFinal) {
[self.audioEngine stop];
[inputNode removeTapOnBus:0];
self.recognitionTask = nil;
self.recognitionRequest = nil;
self.recordButton.enabled = YES;
[self.recordButton setTitle:@"开始录音" forState:UIControlStateNormal];
}
}];
AVAudioFormat *recordingFormat = [inputNode outputFormatForBus:0];
[inputNode removeTapOnBus:0];
[inputNode installTapOnBus:0 bufferSize:1024 format:recordingFormat block:^(AVAudioPCMBuffer * _Nonnull buffer, AVAudioTime * _Nonnull when) {
if (self.recognitionRequest) {
[self.recognitionRequest appendAudioPCMBuffer:buffer];
}
}];
NSError *error;
[self.audioEngine prepare];
[self.audioEngine startAndReturnError:&error];
self.displayLabel.text = @"正在录音。。。";
}
结束录音
- (void)endRecording {
[self.audioEngine stop];
if (_recognitionRequest) {
[_recognitionRequest endAudio];
}
if (_recognitionTask) {
[_recognitionTask cancel];
_recognitionTask = nil;
}
self.recordButton.enabled = NO;
self.displayLabel.text = @"";
}
绑定的语音识别代理
#pragma mark - SFSpeechRecognizerDelegate
- (void)speechRecognizer:(SFSpeechRecognizer *)speechRecognizer availabilityDidChange:(BOOL)available {
if (available) {
NSLog(@"开始录音");
[self.recordButton setTitle:@"开始录音" forState:UIControlStateNormal];
}else {
NSLog(@"语音识别不可用");
[self.recordButton setTitle:@"语音识别不可用" forState:UIControlStateDisabled];
}
self.recordButton.enabled = available;
}
希望这篇文章对各位小伙伴有所帮助，想要Demo的小伙伴点击这里(注意:语音识别只能进行真机调试)
网友评论