优化按住说话说话体验

local_plugins/azure_speech/ios/azure_speech/Sources/tools/SimpleAudioReceiver.swift

@@ -93,7 +93,7 @@ public class SimpleAudioReceiver: NSObject {
                 let frameCount = buffer.frameLength
                 // print("麦克风启动: \(frameCount)")
                 // print("🎵 音频数据: 帧数=\(frameCount), 声道数据=\(channelData != nil ? "有效" : "无效")")
-                self.writeQueue.put(self.audioBufferToData(buffer))
+                self.writeQueue.put(self.audioBufferToDataOptimized(buffer))
                 // 分析或传输数据...
             }
         } catch {
@@ -360,6 +360,7 @@ private func enableVoiceProcessingIfAvailable(inputNode: AVAudioInputNode) {
             }
         }
     }
+    // 修改 runMicrophoneCapture 方法
     private func runMicrophoneCapture() {
         do {
             print("🎤 开始配置麦克风捕获")
@@ -369,27 +370,24 @@ private func enableVoiceProcessingIfAvailable(inputNode: AVAudioInputNode) {
                 print("🔧 创建新的音频引擎")
             }
             
-            // 如果音频引擎正在运行，先停止
             if audioEngine?.isRunning == true {
                 audioEngine?.stop()
                 print("⏹️ 停止现有音频引擎")
             }
             
             try startCapture { buffer in
-                // 处理实时音频数据
+                // 将音频处理移到专用队列，避免阻塞音频线程
-                let channelData = buffer.floatChannelData?[0]
+                self.conversionQueue.async {
-                let frameCount = buffer.frameLength
+                    let audioData = self.audioBufferToDataOptimized(buffer)
-                // print("麦克风启动: \(frameCount)")
+                    self.writeQueue.put(audioData)
-                // print("🎵 音频数据: 帧数=\(frameCount), 声道数据=\(channelData != nil ? "有效" : "无效")")
+                }
-                self.writeQueue.put(self.audioBufferToData(buffer))
-                // 分析或传输数据...
             }
-              try audioEngine?.start()
+            
+            try audioEngine?.start()
             print("✅ 麦克风捕获配置完成")
             
         } catch {
             print("麦克风启动失败: \(error)")
-            // 【新增】添加详细错误处理
             if let nsError = error as NSError? {
                 print("错误域: \(nsError.domain), 错误代码: \(nsError.code)")
                 print("错误描述: \(nsError.localizedDescription)")
@@ -462,63 +460,46 @@ private func enableVoiceProcessingIfAvailable(inputNode: AVAudioInputNode) {
  * @param buffer 音频缓冲区
  * @return 转换后的音频数据
  */
-private func audioBufferToData(_ buffer: AVAudioPCMBuffer) -> Data {
+/**
+ * 优化的音频缓冲区转换方法
+ * 减少内存分配，提高转换效率
+ */
+// 在类属性中添加
+private var reusableDataBuffer: Data?
+private let bufferReuseQueue = DispatchQueue(label: "buffer.reuse", qos: .userInitiated)
+
+/**
+ * 带缓冲区复用的音频转换方法
+ */
+private func audioBufferToDataOptimized(_ buffer: AVAudioPCMBuffer) -> Data {
     guard let floatChannelData = buffer.floatChannelData else {
         return Data()
     }
     
     let frameLength = Int(buffer.frameLength)
-    let channelCount = Int(buffer.format.channelCount)
+    let requiredSize = frameLength * 2
-    let inputSampleRate = buffer.format.sampleRate
+    
-    let targetSampleRate: Double = 16000
+    // 复用或创建缓冲区
+    if reusableDataBuffer == nil || reusableDataBuffer!.count < requiredSize {
+        reusableDataBuffer = Data(count: requiredSize)
+    }
     
-    // 统一处理：只使用第一个声道（单声道），确保一致性
+    var outputData = reusableDataBuffer!
     let firstChannelData = floatChannelData[0]
     
-    // 如果采样率不是16kHz，进行降采样
+    outputData.withUnsafeMutableBytes { rawBytes in
-    if inputSampleRate != targetSampleRate {
+        let int16Ptr = rawBytes.bindMemory(to: Int16.self)
-        let ratio = inputSampleRate / targetSampleRate
-        let outputFrameCount = Int(Double(frameLength) / ratio)
-        
-        var outputData = Data()
-        outputData.reserveCapacity(outputFrameCount * 2) // 16位 = 2字节
-        
-        for outputIndex in 0..<outputFrameCount {
-            let inputIndex = Int(Double(outputIndex) * ratio)
-            if inputIndex < frameLength {
-                let floatSample = firstChannelData[inputIndex]
-                
-                // 将Float32 (-1.0 to 1.0) 转换为Int16 (-32768 to 32767)
-                let clampedSample = max(-1.0, min(1.0, floatSample))
-                let int16Sample = Int16(clampedSample * 32767.0)
-                
-                // 转换为小端字节序
-                withUnsafeBytes(of: int16Sample.littleEndian) { bytes in
-                    outputData.append(contentsOf: bytes)
-                }
-            }
-        }
-        
-        print("音频降采样: \(inputSampleRate)Hz -> \(targetSampleRate)Hz, 帧数: \(frameLength) -> \(outputFrameCount)")
-        return outputData
-    } else {
-        // 采样率已经是16kHz，直接转换格式（只处理第一个声道）
-        var data = Data()
-        data.reserveCapacity(frameLength * 2) // 单声道，16位
         
         for frame in 0..<frameLength {
             let floatSample = firstChannelData[frame]
             let clampedSample = max(-1.0, min(1.0, floatSample))
             let int16Sample = Int16(clampedSample * 32767.0)
-            
+            int16Ptr[frame] = int16Sample.littleEndian
-            withUnsafeBytes(of: int16Sample.littleEndian) { bytes in
-                data.append(contentsOf: bytes)
-            }
         }
-        
-        print("音频格式转换: \(inputSampleRate)Hz, 帧数: \(frameLength), 输出大小: \(data.count) 字节")
-        return data
     }
+    
+    // 返回实际使用的数据部分
+    return outputData.prefix(requiredSize)
 }
 
 /**
@@ -561,13 +542,15 @@ private func convertAudioBuffer(_ buffer: AVAudioPCMBuffer, to targetFormat: AVA
  * @throws 音频会话配置失败时抛出错误
  */
 private func setupAudioSession() throws {
-    try audioSession.setCategory(.playAndRecord, mode: .default, options: [.defaultToSpeaker, .allowBluetooth,.mixWithOthers])
+    // 添加 .mixWithOthers 选项以支持同时播放和录音
+    try audioSession.setCategory(.playAndRecord, 
+                                mode: .default, 
+                                options: [.defaultToSpeaker, .allowBluetooth, .mixWithOthers])
+    
+    // 设置较低的音频延迟以减少卡顿
+    try audioSession.setPreferredIOBufferDuration(0.005) // 5ms
+    
     try audioSession.setActive(true)
-
-    // // 设置首选的音频参数 - 修改为16000Hz以匹配Azure Speech要求
-    // try audioSession.setPreferredSampleRate(16000)
-    // // 设置输入声道数
-    // try audioSession.setPreferredInputNumberOfChannels(1)
     
     print("音频会话配置成功: 采样率=\(audioSession.sampleRate)Hz, 输入声道=\(audioSession.inputNumberOfChannels)")
 }