iOS应用BLE接收PCM音频流播放有杂音，求解决方案及优化建议

核心问题排查与修复 #

1. 音频会话配置错误 #

你初始化里误用了AVCaptureSession（这是音视频捕获专用），音频播放必须用AVAudioSession并配置正确类别：

// 替换原AVCaptureSession为系统共享音频会话
private let audioSession = AVAudioSession.sharedInstance()

init() {
    self.engine = AVAudioEngine()
    self.playerNode = AVAudioPlayerNode()
    
    // 配置音频会话为播放模式
    do {
        try audioSession.setCategory(.playback, mode: .default)
        try audioSession.setActive(true)
    } catch {
        print("音频会话配置失败: \(error)")
    }
    
    self.engine.attach(self.playerNode)
    // 统一使用预定义的format，避免格式不匹配
    engine.connect(self.playerNode, to: engine.outputNode, format: format)
    engine.prepare()
    try! engine.start()
    self.playerNode.play()
}

2. 格式一致性问题 #

初始化连接节点时你混用了standardFormatWithSampleRate和自定义的pcmFormatInt16格式，隐式格式转换会导致音频数据错乱，必须全程使用同一格式。

3. 字节序解析错误 #

你强制用littleEndian转换原始数据，但要确认BLE传来的数据流字节序是否真为小端。如果原始数据是主机字节序，这样转换会直接导致杂音，先去掉字节序转换测试：

let i16array = decoded.withUnsafeBytes {
    Array(UnsafeBufferPointer<Int16>(start: $0.baseAddress!.assumingMemoryBound(to: Int16.self), count: decoded.count / MemoryLayout<Int16>.stride))
}
voicePlayer.play(buffer: i16array)

4. 缓冲区调度连续性问题 #

BLE传输存在延迟、丢包风险，直接调度单个小缓冲区会产生播放间隙杂音。建议维护缓冲区队列，保证连续播放：

private let bufferQueue = DispatchQueue(label: "com.yourapp.audio.buffer")
private var pendingBuffers: [AVAudioPCMBuffer] = []

func play(buffer: [Int16]) {
    bufferQueue.async {
        let frameLength = buffer.count
        guard let audioBuffer = AVAudioPCMBuffer(pcmFormat: self.format, frameCapacity: AVAudioFrameCount(frameLength)) else { return }
        
        audioBuffer.frameLength = AVAudioFrameCount(frameLength)
        let dataSize = buffer.count * MemoryLayout<Int16>.stride
        memcpy(audioBuffer.mutableAudioBufferList.pointee.mBuffers.mData, buffer, dataSize)
        
        self.pendingBuffers.append(audioBuffer)
        
        // 无待播放缓冲区时启动连续调度
        if self.playerNode.numberOfPendingBuffers == 0 {
            self.scheduleNextBuffer()
        }
    }
}

private func scheduleNextBuffer() {
    guard let buffer = pendingBuffers.first else { return }
    pendingBuffers.removeFirst()
    
    playerNode.scheduleBuffer(buffer, completionHandler: { [weak self] in
        self?.scheduleNextBuffer()
    })
}

最优播放方式建议 #

优先选择AVAudioEngine，它提供模块化音频节点链，后续添加滤镜（均衡、降噪、滤波）无需操作底层音频单元，维护成本远低于RemoteIO方案。

带音频滤镜的完整代码示例 #

以下是集成低通滤波器（过滤高频杂音）的完整实现：

import AVFoundation

class VoicePlayer {
    private let audioSession = AVAudioSession.sharedInstance()
    private let engine = AVAudioEngine()
    private let playerNode = AVAudioPlayerNode()
    private let lowPassFilter = AVAudioUnitEQ(numberOfBands: 1)
    private let format = AVAudioFormat(commonFormat: AVAudioCommonFormat.pcmFormatInt16, sampleRate: 8000.0, channels: 1, interleaved: false)!
    private let bufferQueue = DispatchQueue(label: "com.yourapp.audio.buffer")
    private var pendingBuffers: [AVAudioPCMBuffer] = []
    
    init() {
        // 配置低通滤波器，8kHz采样率下截止3kHz过滤高频杂音
        let band = lowPassFilter.bands[0]
        band.filterType = .lowPass
        band.frequency = 3000
        band.bandwidth = 1.0
        band.gain = 0
        
        // 构建节点链：playerNode -> 低通滤波器 -> 输出节点
        engine.attach(playerNode)
        engine.attach(lowPassFilter)
        engine.connect(playerNode, to: lowPassFilter, format: format)
        engine.connect(lowPassFilter, to: engine.outputNode, format: format)
        
        // 激活音频会话
        do {
            try audioSession.setCategory(.playback, mode: .default)
            try audioSession.setActive(true)
        } catch {
            print("音频会话配置失败: \(error)")
        }
        
        // 启动音频引擎
        engine.prepare()
        do {
            try engine.start()
        } catch {
            print("AudioEngine启动失败: \(error)")
        }
        playerNode.play()
    }
    
    func play(buffer: [Int16]) {
        bufferQueue.async {
            let frameLength = buffer.count
            guard let audioBuffer = AVAudioPCMBuffer(pcmFormat: self.format, frameCapacity: AVAudioFrameCount(frameLength)) else { return }
            
            audioBuffer.frameLength = AVAudioFrameCount(frameLength)
            let dataSize = buffer.count * MemoryLayout<Int16>.stride
            memcpy(audioBuffer.mutableAudioBufferList.pointee.mBuffers.mData, buffer, dataSize)
            
            self.pendingBuffers.append(audioBuffer)
            
            if self.playerNode.numberOfPendingBuffers == 0 {
                self.scheduleNextBuffer()
            }
        }
    }
    
    private func scheduleNextBuffer() {
        guard let buffer = pendingBuffers.first else { return }
        pendingBuffers.removeFirst()
        
        playerNode.scheduleBuffer(buffer, completionHandler: { [weak self] in
            self?.scheduleNextBuffer()
        })
    }
    
    deinit {
        playerNode.stop()
        engine.stop()
        do {
            try audioSession.setActive(false)
        } catch {
            print("音频会话停止失败: \(error)")
        }
    }
}

// 使用示例
// let voicePlayer = VoicePlayer()
// 接收BLE数据时解析播放：
// let decoded: Data = ... // BLE传来的原始音频数据
// let i16array = decoded.withUnsafeBytes {
//     Array(UnsafeBufferPointer<Int16>(start: $0.baseAddress!.assumingMemoryBound(to: Int16.self), count: decoded.count / MemoryLayout<Int16>.stride))
// }
// voicePlayer.play(buffer: i16array)

内容的提问来源于stack exchange，提问作者Dhruvi