ios ml 深度学习做字幕

1:使用前需要导入头文件

importSpeech

2:设置请求权限

//请求权限
        SFSpeechRecognizer.requestAuthorization { (state) in
           print(state)
        }

3:配置info.plist

Privacy-Speech Recognition Usage Description   -> “xxx”

4:初始化识别器

//初始化一个识别器

let recognizer = SFSpeechRecognizer(locale: Locale(identifier: "zh_CN” ))

代码

   assert(status == .authorized)
            DispatchQueue.main.async {
            }
            
            //path
            let url = Bundle.main.url(forResource: "video", withExtension: "mp4")
            
            //初始化一个识别请求
            let request = SFSpeechURLRecognitionRequest(url: url!)
            
            //开始一个识别请求
            self.speechRecognizer.recognitionTask(with: request, resultHandler: { (result, error) in
                if error != nil{
                    print("识别错误")
                    return
                }
                else{
                    print(result?.bestTranscription.formattedString)
                }
            })
        }

同步显示字幕

进度条： 参考代码：https://developer.apple.com/library/content/samplecode/AudioTapProcessor/Introduction/Intro.html#//apple_ref/doc/uid/DTS40012324-Intro-DontLinkElementID_2

convert AudioBufferList -> AVAudioPCMBuffer

代码地址：

https://github.com/ssyyjj88/VideoRecognition

未完待续