在Swift 3内部命令行程序中如何执行“说”？

Question

What should I do in this case with the following code?

func convertToM4A(filename: String, voice: String) -> Bool {

    let full_string = speaking_queue?.joined(separator: " ")
    let command_string: [String] = [/"-v \"\(voice)\"",*/ "--progress", "--output-file=\"\(filename)\"","-i", " \"\(full_string!)\""]
    print(command_string)

    /
    let DocumentsDirectory = FileManager().homeDirectory(forUser: "shyamalchandra")
    print((DocumentsDirectory?.absoluteString)!)
    */

    let task = Process()

    task.launchPath = "/usr/bin/say"
    task.arguments = command_string
    let pipe = Pipe()
    task.standardOutput = pipe
    task.launch()

    let data = pipe.fileHandleForReading.readDataToEndOfFile()
    let output: String? = String(data: data, encoding: String.Encoding.utf8)
    task.waitUntilExit()

    if let output = output {
        if !output.isEmpty {
            print(output.trimmingCharacters(in: .whitespacesAndNewlines))
        }
    }

    return true
}

At run-time, it complains about the TERM environment being not set and furthermore, doesn't write the file to disk. What to do?

Answer 1

The main error is how you build the argument array. The given arguments are passed directly to the process. Process does not use the shell to interpret the arguments, therefore you must not enclose them in quotation marks.

Another problem is that the "-i" (interactive) option cannot be used when writing to a file.

So your code should look like this:

func convertToM4A(filename: String, voice: String) -> Bool {

    let fullString = "Hello world"

    let task = Process()
    task.launchPath = "/usr/bin/say"
    task.arguments = [ "-v", voice, "-o", filename, fullString]

    task.launch()
    task.waitUntilExit()

    return true
}

The "--progress" option causes a progress meter to be displayed on standard error. If you want to display that then you would have to read asynchronously from standard error.

Answer 2

If you're writing a native Mac app and want to record synthesized speech to an audio file, don't go trying to wrap a shell command — there's native API for that. NSSpeechSynthesizer is the macOS API for text-to-speech in general, and it has a method startSpeaking(_:to:) that records output to an audio file.

This API outputs to an AIFF file, but there are numerous APIs you can use to convert/encode that to M4A: AVAssetReader / AVAssetWriter , AVAudioFile , lower-level CoreAudio C APIs, etc.

(Generally, if you're writing a native Mac program and there's something you want to do, check to see if there's an API for it before you go trying to wrap a shell command. Usually those shell commands are using the same API, so you're just punishing yourself with all the indirection, I/O parsing, etc.)

Yes, NSSpeechSynthesizer is an AppKit API, but you can use it in a command line tool.

Answer 3

Take a look at this lib, I've used it before and it is very capable of running shell script. With that then you can use the "say" command and send in some arguments. https://github.com/kareman/SwiftShell

You could try it this way for instance

import SwiftShell

try runAndPrint("say", "Hello world", 4, "arguments")

let array = ["Hello world", "we", "are"]
try runAndPrint("say", array, array.count + 2, "arguments")