iOS语音转文字实现

目前正在搞一个IM的APP，内部好友之间可以发送语音，需要长按实现语音转文字的功能，之前使用的是阿里的NUI.framework,但是这个破玩意，经常出现转出来的文字重复，即使了多声道控制都无法处理掉，体验太差。没办法，就决定替换为apple自己的实现，毕竟siri那么强大！此实现包含本地音频及远程音频，你只需要按时数据model保存对应的path即可，内部会自动识别。

现在来看看实现条件：

在 Info.plist里面添加两个键值对：

1、Privacy - Speech Recognition Usage Description （用于请求语音识别）

2、 Privacy - Microphone Usage Description（用于请求麦克风语音输入授权）。

并给出相应的文字描述。

导入库文件：

#import <Speech/Speech.h>

以下为实现头文件及逻辑文件：

头文件： NSVoice2Text.h

//
//  NSVoice2Text.h
//  语音转文字
//
//  Created by wise on 2021/10/13.
//#import <Foundation/Foundation.h>
#import <Speech/Speech.h>NS_ASSUME_NONNULL_BEGINtypedef NS_ENUM(NSUInteger, NSVoice2TextAuthorationStatus) {NSVoice2TextAuthorizationStatusNotDetermined,  //语音识别未授权NSVoice2TextAuthorizationStatusDenied,         //用户拒绝使用语音识别NSVoice2TextAuthorizationStatusRestricted,     //语音识别在这台设备上受到限制NSVoice2TextAuthorizationStatusAuthorized,     //可以语音识别
};@interface NSVoiceModel : NSObject
@property (nonatomic,copy) NSString *path;@property (nonatomic,assign) NSInteger taskId;@property (nonatomic,assign) BOOL isRunning;@property (nonatomic,assign) BOOL isInQueue;
@end@interface NSVoice2TextFinal : NSObject
@property (nonatomic,copy) NSString *value;@property (nonatomic,assign) NSInteger taskId;@property (nonatomic,copy) NSError * __nullable error;
@end@interface NSVoice2Text : NSObject+ (BOOL) isRunning;//权限
+ (void)voice2TextRequestAuthorationStatus:(void (^)(NSVoice2TextAuthorationStatus status))requestBlock;+ (void)voice2TextGotter:(NSArray <NSVoiceModel *>*)glist runningModelBlock:(void (^__nullable)(NSVoiceModel *amodel))runningModelBlock resultsBlock:(void (^)(NSVoice2TextFinal *finalValue))resultsBlock rtaget:(id)rtaget;@endNS_ASSUME_NONNULL_END

实现文件：NSVoice2Text.m

//
//  NSVoice2Text.m
//  语音转文字
//
//  Created by wise on 2021/10/13.
//#import "NSVoice2Text.h"
#import "NSMutableTaskQueue.h"typedef void (^VoiceConversionResultsBlock) (NSVoice2TextFinal *finalValue);@interface NSVoiceModel ()
@property (nonatomic,weak) id taskTarget;@property (nonatomic, copy) VoiceConversionResultsBlock voiceConversionBlock;@property (nonatomic, copy) void (^voiceConversionRunningBlock)(NSVoiceModel *md);
@end@implementation NSVoiceModel@end@implementation NSVoice2TextFinal
@endstatic NSVoice2Text *v2text = nil;@interface NSVoice2Text ()<SFSpeechRecognizerDelegate>
{NSMutableArray <NSVoiceModel *>* taskList;
}@property (nonatomic, assign) NSVoice2TextAuthorationStatus authorationStatus;@property(nonatomic,strong)SFSpeechRecognizer *speechRecognizer;//语音识别器@end@implementation NSVoice2Text
- (instancetype)init
{self = [super init];if (self){taskList = [NSMutableArray arrayWithCapacity:0];}return self;
}+ (instancetype)shareInstance
{if (!v2text){v2text = [[NSVoice2Text alloc] init];}return v2text;
}+ (void)releaseInstance
{if (v2text){v2text = nil;}
}- (SFSpeechRecognizer *)speechRecognizer
{if (_speechRecognizer == nil) {NSLocale *cale = [[NSLocale alloc]initWithLocaleIdentifier:@"zh-CN"];_speechRecognizer = [[SFSpeechRecognizer alloc]initWithLocale:cale];_speechRecognizer.delegate = self;}return _speechRecognizer;
}+ (BOOL) isRunning
{NSVoiceModel *md = [[NSVoice2Text shareInstance]->taskList firstObject];return md.isRunning;
}- (void)resume
{NSVoiceModel *md = [self->taskList firstObject];if (md && !md.isInQueue){md.isInQueue = YES;if (md.voiceConversionRunningBlock){md.voiceConversionRunningBlock(md);}if (md.path && md.path > 0 && !md.isRunning){md.isRunning = YES;NSString *text = @"^(http|https)+.*";NSPredicate *regextest = [NSPredicate predicateWithFormat:@"SELF MATCHES %@", text];BOOL flag = [regextest evaluateWithObject:md.path];if (flag){[self startVoiceConversionWithURL:md.path];}else{[self startVoiceConversionWithFilePath:md.path];}}else{NSVoice2TextFinal *el = [[NSVoice2TextFinal alloc] init];el.taskId = -1;el.error = [NSError errorWithDomain:@"语音路径错误或为空" code:404 userInfo:nil];md.voiceConversionBlock(el);}}
}- (void)addItToTask:(NSVoiceModel *)md
{__block BOOL contained = NO;[taskList enumerateObjectsUsingBlock:^(NSVoiceModel * _Nonnull obj, NSUInteger idx, BOOL * _Nonnull stop) {if (obj.taskId == md.taskId){contained = YES;*stop = YES;}}];if (!contained){[taskList addObject:md];}
}+ (void)voice2TextRequestAuthorationStatus:(void (^)(NSVoice2TextAuthorationStatus status))requestBlock
{//发送语音认证请求(首先要判断设备是否支持语音识别功能)[SFSpeechRecognizer requestAuthorization:^(SFSpeechRecognizerAuthorizationStatus status){[[NSVoice2Text shareInstance] setAuthorationStatus:status];requestBlock(status);}];
}+ (void)voice2TextGotter:(NSArray <NSVoiceModel *>*)glist runningModelBlock:(void (^__nullable)(NSVoiceModel *amodel))runningModelBlock resultsBlock:(void (^)(NSVoice2TextFinal *finalValue))resultsBlock rtaget:(id)rtaget
{[glist enumerateObjectsUsingBlock:^(NSVoiceModel * _Nonnull obj, NSUInteger idx, BOOL * _Nonnull stop){[obj setVoiceConversionRunningBlock:runningModelBlock];[obj setVoiceConversionBlock:resultsBlock];[obj setTaskTarget:rtaget];[[NSVoice2Text shareInstance] addItToTask:obj];}];[[NSVoice2Text shareInstance] resume];
}- (void)startVoiceConversionWithFilePath:(NSString *)path
{[self startVoiceConversion:[NSURL fileURLWithPath:path]];
}- (void)startVoiceConversionWithURL:(NSString *)url
{[self startVoiceConversion:[NSURL URLWithString:url]];
}#pragma mark - private methods
///开始转换
- (void)startVoiceConversion:(NSURL *)url
{__weak typeof(taskList) weakTaskList = taskList;__weak typeof(self) this = self;SFSpeechURLRecognitionRequest *recognitionRequest = [[SFSpeechURLRecognitionRequest alloc]initWithURL:url];NSLocale *cale = [[NSLocale alloc]initWithLocaleIdentifier:@"zh-CN"];SFSpeechRecognizer *sp = [[SFSpeechRecognizer alloc]initWithLocale:cale];NSOperationQueue *otherQuene = [[NSOperationQueue alloc]init];[sp setQueue:otherQuene];[sp recognitionTaskWithRequest:recognitionRequest resultHandler:^(SFSpeechRecognitionResult * _Nullable result, NSError *  error){NSVoiceModel *md = [weakTaskList firstObject];if (md.taskTarget){if (!error){if (result){BOOL isFinal = [result isFinal];//是否结束if (isFinal){NSString *str = [[result bestTranscription]formattedString];NSVoice2TextFinal *el = [[NSVoice2TextFinal alloc] init];el.taskId = md.taskId;el.error = nil;el.value = str;md.voiceConversionBlock(el);[weakTaskList removeObject:md];[this resume];}}else{NSVoice2TextFinal *el = [[NSVoice2TextFinal alloc] init];el.taskId = md.taskId;el.error = error;md.voiceConversionBlock(el);[weakTaskList removeObject:md];[this resume];}}else{NSVoice2TextFinal *el = [[NSVoice2TextFinal alloc] init];el.taskId = md.taskId;el.error = error;md.voiceConversionBlock(el);[weakTaskList removeObject:md];[this resume];}}else{[weakTaskList removeObject:md];[this resume];}}];
}
@end

此实现内部已经实现了队列转文字功能，你只需要随时传入数据模型即可。

代码分析：

1、权限请求

+ (void)voice2TextRequestAuthorationStatus:(void (^)(NSVoice2TextAuthorationStatus status))requestBlock;

用于请求隐私权限，只有用户同意后方可使用此功能。否则无法使用此功能。

2、传入音频文件路径

+ (void)voice2TextGotter:(NSArray <NSVoiceModel *>*)glist runningModelBlock:(void (^__nullable)(NSVoiceModel *amodel))runningModelBlock resultsBlock:(void (^)(NSVoice2TextFinal *finalValue))resultsBlock rtaget:(id)rtaget

音频以数据模型NSVoiceModel传入，将你的音频文件与此模型实现映射关系,taskID用于实现绑定,参考头文件的定义及实现。

2.1 runningModelBlock，因为支持队列事务，所以，当前正在处理哪条，则会对外输出此条。页面上可以此显示"正在转换中"文字

2.2 resultsBlock，转换结果文字，以NSVoice2TextFinal对外输出，你只需要处理好这里面的逻辑好可。

3、完整使用：

[NSVoice2Text voice2TextRequestAuthorationStatus:^(NSVoice2TextAuthorationStatus status){if (status == NSVoice2TextAuthorizationStatusAuthorized){NSVoiceModel *md = [[NSVoiceModel alloc] init];[md setTaskId:[bmodel.messageId integerValue]];[md setPath:bmodel.audioFilePath];[NSVoice2Text voice2TextGotter:@[md] runningModelBlock:^(NSVoiceModel * _Nonnull amodel){NSString *taskId = intToStr(amodel.taskId);//通过taskId找到对应的处理的UI，显示"正在转换中"}resultsBlock:^(NSVoice2TextFinal * _Nonnull finalValue){if (!finalValue.error){NSString *taskId = intToStr(finalValue.taskId);NSString *trTexgt = [finalValue value];//通过taskId找到对应的处理的UI，转换完成，得到转换后的文字}else{NSString *taskId = intToStr(finalValue.taskId);//此taskId对应的语音转换失败，亦可找到对应的UI，显示"转换失败"等文字}}];}else{[weakSelf showToastMessageThenHide:@"未授权使用语音识别功能"];}}rtaget:weakSelf];

iOS语音转文字实现相关推荐

iOS 语音播放文字内容--制作简易听书软件(AVSpeechSynthesizer)
iOS 语音播放文字内容--制作简易听书软件(AVSpeechSynthesizer) 字数46 阅读731 评论8 喜欢50 Collection/Bookmark/Share for width ...
iOS 语音读文字so easy
#import <AVFoundation/AVFoundation.h> // 点击事件里读文字 -(void)touchesBegan:(NSSet<UITouch *> ...
iOS开发之语音朗读文字
本文使用AVSpeechSynthesizer和AVSpeechUtterance两个类来完成语音朗读文字功能. 需要: 为项目添加AVFoundation框架导入头文件import AVFound ...
android ios语音转码,手机如何将语音转文字？这几种方法真简单，安卓苹果通用...
原标题:手机如何将语音转文字?这几种方法真简单,安卓苹果通用手机可以将语音转文字吗? 当然可以啦,方法还不止一种,下面就简单给大家介绍几种,安卓苹果手机通用的方法吧. 1.输入法-实时语音转文字手 ...
文字转wav_这6款超良心语音转文字工具，真让人省心！
今天的文章,小叶子想帮大家解决「语音转文字」的痛点. 工作学习中,我们总会遇上语音转文字的需求,例如工作会议布置大量任务.期末最后一节课划重点,语速太快来不及记录,只能录音. 但录音一时爽,听时「火葬 ...
普通话转粤语_语音转文字评测：几款语音转文字app，你了解多少？
语音转文字有必要吗?能用在哪里?这是大多数人对于语记类app的疑问所在,今天为大家简单介绍一下几款实用的语记app以及简单的应用描述. 1. 讯飞语记讯飞语记是讯飞旗下的语音转文字产品,如我们所知, ...
三大运营商回复 4G 降速；微信上线语音转文字功能；IntelliJ IDEA 2019.2.1 发布 | 极客头条...
快来收听极客头条音频版吧,智能播报由标贝科技提供技术支持. 「CSDN 极客头条」,是从 CSDN 网站延伸至官方微信公众号的特别栏目,专注于一天业界事报道.风里雨里,我们将每天为朋友们,播报最新鲜有 ...
android语音输入文字,盘点好用的语音输入APP，懒得打字的时候就说话吧！
原标题:盘点好用的语音输入APP,懒得打字的时候就说话吧! 本文为「智活范」原创作品,欢迎关注我们! 上次推完好用的录音APP后,立刻就有萌友来问了,能不能直接录音转文字呢,这样说话就能生成文字,多省 ...
AR眼镜语音转文字实测！效果像开了弹幕，对话记录可保存回溯
明敏发自凹非寺量子位 | 公众号 QbitAI AR眼镜字幕功能效果到底咋样? 实测来了! 不光语音能实时转成文字,还能分辨说话人主体,甚至还能把文字记录都保存好,方便以后回溯. 今年,一家来自 ...
使用deepspeech.pytorch项目对中文普通话数据集进行语音转文字
目录介绍注意事项实验过程 thchs30 aishell Primewords Chinese Corpus Set 1 Free ST Chinese Mandarin Corpus Aida ...

iOS语音转文字实现

iOS语音转文字实现相关推荐

最新文章

热门文章