Text-to-speech for iOS (#443)

.gitignore

@@ -67,6 +67,13 @@ swift-api-examples/k2fsa-*
 run-*.sh
 two-pass-*.sh
 build-*
+
+## User settings
+xcuserdata/
+
+## Xcode 8 and earlier
+*.xcscmblueprint
+*.xccheckout
 vits-vctk
 vits-zh-aishell3
 jslint.mjs

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+	<key>IDEDidComputeMac32BitWarning</key>
+	<true/>
+</dict>
+</plist>

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/Assets.xcassets/AccentColor.colorset/Contents.json

@@ -0,0 +1,11 @@
+{
+  "colors" : [
+    {
+      "idiom" : "universal"
+    }
+  ],
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/Assets.xcassets/AppIcon.appiconset/Contents.json

@@ -0,0 +1,13 @@
+{
+  "images" : [
+    {
+      "idiom" : "universal",
+      "platform" : "ios",
+      "size" : "1024x1024"
+    }
+  ],
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/Assets.xcassets/Contents.json

@@ -0,0 +1,6 @@
+{
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/ContentView.swift

@@ -0,0 +1,95 @@
+//
+//  ContentView.swift
+//  SherpaOnnxTts
+//
+//  Created by fangjun on 2023/11/23.
+//
+// Speech-to-text with Next-gen Kaldi on iOS without Internet connection
+
+import SwiftUI
+import AVFoundation
+
+struct ContentView: View {
+    @State private var sid = "0"
+    @State private var speed = 1.0
+    @State private var text = ""
+    @State private var showAlert = false
+    @State var filename: URL = NSURL() as URL
+    @State var audioPlayer: AVAudioPlayer!
+
+    private var tts = createOfflineTts()
+
+    var body: some View {
+
+        VStack(alignment: .leading) {
+            HStack {
+                Spacer()
+                Text("Next-gen Kaldi: TTS").font(.title)
+                Spacer()
+            }
+            HStack{
+                Text("Speaker ID")
+                TextField("Please input a speaker ID", text: $sid).textFieldStyle(.roundedBorder)
+                    .keyboardType(.numberPad)
+            }
+            HStack{
+                Text("Speed \(String(format: "%.1f", speed))")
+                    .padding(.trailing)
+                Slider(value: $speed, in: 0.5...2.0, step: 0.1) {
+                    Text("Speech speed")
+                }
+            }
+
+            Text("Please input your text below").padding([.trailing, .top, .bottom])
+
+            TextEditor(text: $text)
+                .font(.body)
+                .opacity(self.text.isEmpty ? 0.25 : 1)
+                .disableAutocorrection(true)
+                .border(Color.black)
+
+            Spacer()
+            HStack {
+                Spacer()
+                Button(action: {
+                    let speakerId = Int(self.sid) ?? 0
+                    let t = self.text.trimmingCharacters(in: .whitespacesAndNewlines)
+                    if t.isEmpty {
+                        self.showAlert = true
+                        return
+                    }
+
+                    let audio = tts.generate(text: t, sid: speakerId, speed: Float(self.speed))
+                    if self.filename.absoluteString.isEmpty {
+                        let tempDirectoryURL = NSURL.fileURL(withPath: NSTemporaryDirectory(), isDirectory: true)
+                        self.filename = tempDirectoryURL.appendingPathComponent("test.wav")
+                    }
+
+                    let ret = audio.save(filename: filename.path)
+
+                    self.audioPlayer = try! AVAudioPlayer(contentsOf: filename)
+                    self.audioPlayer.play()
+                }) {
+                    Text("Generate")
+                }.alert(isPresented: $showAlert) {
+                    Alert(title: Text("Empty text"), message: Text("Please input your text before clicking the Generate button"))
+                }
+                Spacer()
+                Button (action: {
+                    self.audioPlayer.play()
+                }) {
+                    Text("Play")
+                }.disabled(filename.absoluteString.isEmpty)
+                Spacer()
+            }
+            Spacer()
+        }
+        .padding()
+    }
+}
+
+struct ContentView_Previews: PreviewProvider {
+    static var previews: some View {
+        ContentView()
+    }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/Preview Content/Preview Assets.xcassets/Contents.json

@@ -0,0 +1,6 @@
+{
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/SherpaOnnxTtsApp.swift

@@ -0,0 +1,17 @@
+//
+//  SherpaOnnxTtsApp.swift
+//  SherpaOnnxTts
+//
+//  Created by fangjun on 2023/11/23.
+//
+
+import SwiftUI
+
+@main
+struct SherpaOnnxTtsApp: App {
+    var body: some Scene {
+        WindowGroup {
+            ContentView()
+        }
+    }
+}

ios-swiftui/SherpaOnnxTts/SherpaOnnxTts/ViewModel.swift

@@ -0,0 +1,68 @@
+//
+//  ViewModel.swift
+//  SherpaOnnxTts
+//
+//  Created by fangjun on 2023/11/23.
+//
+
+import Foundation
+
+func getResource(_ forResource: String, _ ofType: String) -> String {
+  let path = Bundle.main.path(forResource: forResource, ofType: ofType)
+  precondition(
+    path != nil,
+    "\(forResource).\(ofType) does not exist!\n" + "Remember to change \n"
+      + "  Build Phases -> Copy Bundle Resources\n" + "to add it!"
+  )
+  return path!
+}
+
+/// Please refer to
+/// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/index.html
+/// to download pre-trained models
+
+func getTtsForVCTK() -> SherpaOnnxOfflineTtsWrapper {
+  // See the following link
+  // https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/vits.html#vctk-english-multi-speaker-109-speakers
+
+  // vits-vctk.onnx
+  let model = getResource("vits-vctk", "onnx")
+
+  // lexicon.txt
+  let lexicon = getResource("lexicon", "txt")
+
+  // tokens.txt
+  let tokens = getResource("tokens", "txt")
+
+  let vits = sherpaOnnxOfflineTtsVitsModelConfig(model: model, lexicon: lexicon, tokens: tokens)
+  let modelConfig = sherpaOnnxOfflineTtsModelConfig(vits: vits)
+  var config = sherpaOnnxOfflineTtsConfig(model: modelConfig)
+  return SherpaOnnxOfflineTtsWrapper(config: &config)
+}
+
+func getTtsForAishell3() -> SherpaOnnxOfflineTtsWrapper {
+  // See the following link
+  // https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/vits.html#vits-model-aishell3
+
+  // vits-vctk.onnx
+  let model = getResource("vits-aishell3", "onnx")
+
+  // lexicon.txt
+  let lexicon = getResource("lexicon", "txt")
+
+  // tokens.txt
+  let tokens = getResource("tokens", "txt")
+
+  let vits = sherpaOnnxOfflineTtsVitsModelConfig(model: model, lexicon: lexicon, tokens: tokens)
+  let modelConfig = sherpaOnnxOfflineTtsModelConfig(vits: vits)
+  var config = sherpaOnnxOfflineTtsConfig(model: modelConfig)
+  return SherpaOnnxOfflineTtsWrapper(config: &config)
+}
+
+func createOfflineTts() -> SherpaOnnxOfflineTtsWrapper {
+  return getTtsForVCTK()
+
+  // return getTtsForAishell3()
+
+  // please add more models on need by following the above two examples
+}

swift-api-examples/SherpaOnnx.swift

@@ -650,8 +650,8 @@ class SherpaOnnxGeneratedAudioWrapper {
     }
   }
 
-  func save(filename: String) {
-    SherpaOnnxWriteWave(audio.pointee.samples, n, sampleRate, toCPointer(filename))
+  func save(filename: String) -> Int32 {
+    return SherpaOnnxWriteWave(audio.pointee.samples, n, sampleRate, toCPointer(filename))
   }
 }