同步口譯app的開發過程

 同步口譯app的開發過程

iPhone

即時語音辨識用內建的就很強

    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "zh-TW"))!

//    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "ja-JP"))!

//    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "en-US"))!

//    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "th-TH"))!

//    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "vi-VN"))!

//    private let speechRecognizer = SFSpeechRecognizer(locale: Locale(identifier: "ko-KR"))!

    private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?

    

    private var recognitionTask: SFSpeechRecognitionTask?

    

    private let audioEngine = AVAudioEngine()

        speechRecognizer.delegate = self

        

        // Asynchronously make the authorization request.

        SFSpeechRecognizer.requestAuthorization

  

可以參考xcode中的範例 SpokenWord

然後，翻譯的部分，我先嚐試用 google mlkit的翻譯，但翻譯品質很差，我打算改用google 雲端付費的api來試試，每月有50萬字的免費額度

google mlkit translate的作法

要先引用套件

pod 'GoogleMLKit/Translate', '4.0.0'

然後使用的方法是

var mytr: Translator?

        let options = TranslatorOptions(sourceLanguage: .chinese, targetLanguage: .japanese)

        mytr = Translator.translator(options: options)

        let conditions = ModelDownloadConditions(

            allowsCellularAccess: false,

            allowsBackgroundDownloading: true

        )

        mytr?.downloadModelIfNeeded(with: conditions) { error in

            guard error == nil else { return }

            print("model ok")

            // Model downloaded successfully. Okay to start translating.

        }

                self.mytr?.translate(self.tv_talk.text) { translatedText, error in

                    guard error == nil, let translatedText = translatedText else { return }

                        self.tv_my.text=translatedText

                }

同樣也有範例參考 TranslateExample

下載位置 https://github.com/googlesamples/mlkit