Description: An easy implemenation of audio transcription in swift
Input: Audio file such as .mp3 file
Output: Transcription
Three essential components are: Title, Text space, Button, and Activity spinner.
Title describes the purpose of this app. Title component is very simple. You can use a label in Swift to make it.You can center and horizontal alignment for the title.
Text space is to display the final transcription results.So under your title, add a text field for it.
Button is the main function of this app. If an user presses it, the app will make a transcription. Activity spinner plays a spinner animation to indicate the transcription process. First, the basic setup is also simple. You can use button in swift to make it and add any alignment you like.
Next, we need to add Text space, Activity spinner, and Button in viewcontroller. Right click the button and connect it to viewcontroller.
You could use the ViewController, AppDelegate, and SceneDelegate file I posed in this github.
The following part is to explain some parts of it.
Inside the outlet, we first define three To-do functions.
@IBAction func playButtonPressed(_ sender: Any) {
activitySpinner.isHidden = false # Show activity spinner
activitySpinner.startAnimating()# Start animation of activity spinner
requestSpeechAuth()# Start transcription
}
First function is to show the hidden activitySpinner. Second function is to start animation of activity spinner.
First and second function were already defined by Xcode. So we don't need to do anything.
Third function is to start transcription. This function calls a mp3 file from local file and starts transcription. SFSpeechRecognizer is the main variable. We need to feed it with a path, and it will do the task.
func requestSpeechAuth() {
SFSpeechRecognizer.requestAuthorization { authStatus in
if authStatus == SFSpeechRecognizerAuthorizationStatus.authorized{
if let path = Bundle.main.url(forResource: "test", withExtension: "mp3"){
do{
let sound = try AVAudioPlayer(contentsOf: path)
self.audioPlay = sound
self.audioPlay.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 was an error:\(error)")
} else {
print(result?.bestTranscription.formattedString)
self.transcriptionTextField.text = result?.bestTranscription.formattedString
}
}
}
}
}
}
Finally, when we press this button, activity spinner will start spin and swift can call SFSpeechRecognizer to detect the most likely transcription.
