iOS
API de reconnaissance vocale iOS 10

Discours au texte: reconnaître la parole d'un paquet contenant un enregistrement audio

//import Speech
//import AVFoundation

// create a text field to show speech output
@IBOutlet weak var transcriptionTextField: UITextView!
// we need this audio player to play audio
var audioPlayer: AVAudioPlayer!

override func viewDidLoad()
{
    super.viewDidLoad()
}

// this function is required to stop audio on audio completion otherwise it will play same audio again and again 
func audioPlayerDidFinishPlaying(_ player: AVAudioPlayer, successfully flag: Bool)
{
    player.stop()
}

// this function is required to get a speech recognizer and after that make and request to speech recognizer 
func requestSpeechAuth()
{
    SFSpeechRecognizer.requestAuthorization { authStatus in
        if authStatus == SFSpeechRecognizerAuthorizationStatus.authorized {
            if let path = Bundle.main.url(forResource: "mpthreetest", withExtension: "m4a") {
                do {
                    let sound = try AVAudioPlayer(contentsOf: path)
                    self.audioPlayer = sound
                    self.audioPlayer.delegate = self
                    sound.play()
                }   catch {
                    print("error")
                }
                
                let recognizer = SFSpeechRecognizer()
                let request = SFSpeechURLRecognitionRequest(url:path)
                recognizer?.recognitionTask(with: request) { (result, error) in
                    if let error = error {
                    print("there is a error\(error)")
                    } else {
// here you are printing out the audio output basically showing it on uitext field
                        self.transcriptionTextField.text = result?.bestTranscription.formattedString
                    }
                }
            }
        }
    }
}

// here you are calling requestSpeechAuth function on UIButton press
@IBAction func playButtonPress(_ sender: AnyObject)
{
    requestSpeechAuth()
}