feat: add MediaStreamTrack support to speech recognition

src/speech.test.ts

@@ -10,7 +10,7 @@
  */
 describe('speech function', () => {
   // モックされたイベントリスナーを保存するオブジェクト
   let eventListeners: Record<string, Array<(...args: any[]) => void>> = {}
 
   // モックされたSpeechRecognitionインスタンス
   let mockRecognitionInstance: {
@@ -156,6 +156,91 @@
 
     // Assert: 期待される結果を確認
     expect(mockRecognitionInstance.start).toHaveBeenCalledTimes(1)
+    expect(mockRecognitionInstance.start).toHaveBeenCalledWith()
+  })
+
+  it('When calling start method with MediaStreamTrack, it passes the track to recognition.start', () => {
+    // Arrange: 操作に必要な準備
+    const recognitionObj = speech({})
+    const mockAudioTrack = {
+      kind: 'audio',
+      readyState: 'live',
+    } as MediaStreamTrack
+
+    // Act: 結果を得るために必要な操作
+    recognitionObj.start(mockAudioTrack)
+
+    // Assert: 期待される結果を確認
+    expect(mockRecognitionInstance.start).toHaveBeenCalledTimes(1)
+    expect(mockRecognitionInstance.start).toHaveBeenCalledWith(mockAudioTrack)
+  })
+
+  it('When calling start method with video track, it throws InvalidStateError', () => {
+    // Arrange: 操作に必要な準備
+    const recognitionObj = speech({})
+    const mockVideoTrack = {
+      kind: 'video',
+      readyState: 'live',
+    } as MediaStreamTrack
+
+    // Act & Assert: 操作とアサーションを組み合わせる
+    expect(() => recognitionObj.start(mockVideoTrack)).toThrow(DOMException)
+    expect(() => recognitionObj.start(mockVideoTrack)).toThrow(
+      'The provided MediaStreamTrack must be an audio track'
+    )
+    expect(mockRecognitionInstance.start).not.toHaveBeenCalled()
+  })
+
+  it('When calling start method with ended audio track, it throws InvalidStateError', () => {
+    // Arrange: 操作に必要な準備
+    const recognitionObj = speech({})
+    const mockEndedTrack = {
+      kind: 'audio',
+      readyState: 'ended',
+    } as MediaStreamTrack
+
+    // Act & Assert: 操作とアサーションを組み合わせる
+    expect(() => recognitionObj.start(mockEndedTrack)).toThrow(DOMException)
+    expect(() => recognitionObj.start(mockEndedTrack)).toThrow(
+      'The provided MediaStreamTrack must be in "live" state'
+    )
+    expect(mockRecognitionInstance.start).not.toHaveBeenCalled()
+  })
+
+  it('When audioTrack is provided in options, it uses that track on start', () => {
+    // Arrange: 操作に必要な準備
+    const mockAudioTrack = {
+      kind: 'audio',
+      readyState: 'live',
+    } as MediaStreamTrack
+    const recognitionObj = speech({ audioTrack: mockAudioTrack })
+
+    // Act: 結果を得るために必要な操作
+    recognitionObj.start()
+
+    // Assert: 期待される結果を確認
+    expect(mockRecognitionInstance.start).toHaveBeenCalledTimes(1)
+    expect(mockRecognitionInstance.start).toHaveBeenCalledWith(mockAudioTrack)
+  })
+
+  it('When both options.audioTrack and parameter audioTrack are provided, parameter takes precedence', () => {
+    // Arrange: 操作に必要な準備
+    const optionsTrack = {
+      kind: 'audio',
+      readyState: 'live',
+    } as MediaStreamTrack
+    const parameterTrack = {
+      kind: 'audio',
+      readyState: 'live',
+    } as MediaStreamTrack
+    const recognitionObj = speech({ audioTrack: optionsTrack })
+
+    // Act: 結果を得るために必要な操作
+    recognitionObj.start(parameterTrack)
+
+    // Assert: 期待される結果を確認
+    expect(mockRecognitionInstance.start).toHaveBeenCalledTimes(1)
+    expect(mockRecognitionInstance.start).toHaveBeenCalledWith(parameterTrack)
   })
 
   it('When calling stop method, it stops the recognition', () => {

src/speech.ts

@@ -50,6 +50,12 @@ export interface SpeechOptions {
    * エラー時のコールバック
    */
   onError?: (error: SpeechRecognitionErrorCode) => void
+
+  /**
+   * 音声入力として使用するMediaStreamTrack
+   * 指定しない場合はデフォルトのマイクを使用
+   */
+  audioTrack?: MediaStreamTrack
 }
 
 /**
@@ -111,9 +117,29 @@ export function speech(options: SpeechOptions = {}) {
   return {
     /**
      * 音声認識を開始
+     * @param audioTrack - オプション: 使用する音声トラック
      */
-    start: () => {
-      recognition.start()
+    start: (audioTrack?: MediaStreamTrack) => {
+      const trackToUse = audioTrack || options.audioTrack
+
+      if (trackToUse) {
+        if (trackToUse.kind !== 'audio') {
+          throw new DOMException(
+            'The provided MediaStreamTrack must be an audio track',
+            'InvalidStateError'
+          )
+        }
+        if (trackToUse.readyState !== 'live') {
+          throw new DOMException(
+            'The provided MediaStreamTrack must be in "live" state',
+            'InvalidStateError'
+          )
+        }
+        // @ts-expect-error - Web Speech API仕様の新しいメソッドシグネチャ
+        recognition.start(trackToUse)
+      } else {
+        recognition.start()
+      }
     },
 
     /**