Android平台Codec2编解码实现遇音频失真问题求助

问题根源分析 #

你的代码存在三个核心问题，直接导致音频严重失真：

1. 编码时重复处理同一帧数据：录音缓冲区包含4帧有效数据（640样本/160样本每帧），但你每次编码都传入整个缓冲区，相当于重复编码第一帧4次，导致数据冗余错乱。
2. 用CharArray存储编码字节流：Codec2输出的是原始二进制字节数据，用CharArray存储会引入字符编码转换问题，导致解码时字节值丢失或异常。
3. 解码时字节转换逻辑错误：将CharArray转ByteArray的方式会破坏原始字节结构，比如高位字节被截断，导致解码出的PCM数据完全错误。

修复后的代码实现 #

companion object {
    public const val codec2mode = Codec2.CODEC2_MODE_3200

    private const val sampleRate = 8000
    private const val audioFormat = AudioFormat.ENCODING_PCM_16BIT
    private const val monoStereoIn = AudioFormat.CHANNEL_IN_MONO
    private const val monoStereoOut = AudioFormat.CHANNEL_OUT_MONO
}

private val minBufSize = AudioRecord.getMinBufferSize(sampleRate, monoStereoIn, audioFormat)

private val c2instance = Codec2.create(codec2mode)
private val frameBytesSize = Codec2.getBitsSize(c2instance) // 注意：返回值是每帧编码后的字节数
private val samplesPerFrame = Codec2.getSamplesPerFrame(c2instance)

fun recordData(): ByteArray {
    Timber.i("recordData frameBytes:$frameBytesSize, samplesPerFrame:$samplesPerFrame, minBufSize:$minBufSize")
    val recorder = getRecorder()

    val recorderBuffer = ShortArray(minBufSize)
    val framesPerBuffer = recorderBuffer.size / samplesPerFrame

    Timber.i("startRecording recorderBuffer size:${recorderBuffer.size} framesPerBuffer:$framesPerBuffer")
    recorder.startRecording()

    val start = System.currentTimeMillis()
    val encodedData = mutableListOf<Byte>()

    var it = 50
    while (it > 0) {
        // 必须处理read返回值，确保读取到有效数据
        val readCount = recorder.read(recorderBuffer, 0, recorderBuffer.size)
        if (readCount <= 0) {
            Timber.w("Failed to read audio data")
            it--
            continue
        }

        // 分帧处理每一段样本
        for (i in 0 until framesPerBuffer) {
            val frameStart = i * samplesPerFrame
            // 避免越界：如果读取的样本数不足一整帧则跳过
            if (frameStart + samplesPerFrame > readCount) break
            
            val frameSamples = recorderBuffer.copyOfRange(frameStart, frameStart + samplesPerFrame)
            val encodedFrame = ByteArray(frameBytesSize)
            Codec2.encode(c2instance, frameSamples, encodedFrame)
            encodedData.addAll(encodedFrame.toList())
        }
        it--
    }

    recorder.stop()
    recorder.release()

    Timber.i(
        "encode finished in ${System.currentTimeMillis() - start}ms, data length:${encodedData.size}"
    )

    return encodedData.toByteArray()
}

fun playData(encodedBuffer: ByteArray) {
    Timber.i("playData frameBytes:$frameBytesSize, samplesPerFrame:$samplesPerFrame, minBufSize:$minBufSize")
    val start = System.currentTimeMillis()

    val audioTrack = getAudioTrack().apply { play() }

    var offset = 0
    var it = 0
    while (offset + frameBytesSize <= encodedBuffer.size) {
        // 截取当前帧的编码字节
        val encodedFrame = encodedBuffer.copyOfRange(offset, offset + frameBytesSize)
        val playbackAudioBuffer = ShortArray(samplesPerFrame)
        
        Codec2.decode(c2instance, playbackAudioBuffer, encodedFrame)
        audioTrack.write(playbackAudioBuffer, 0, playbackAudioBuffer.size)

        offset += frameBytesSize
        it++
    }

    audioTrack.stop()
    audioTrack.release()

    Timber.i(
        "decode and play finished in ${System.currentTimeMillis() - start}ms, iterations:$it"
    )
}

// 在合适的生命周期方法中释放Codec2实例，避免内存泄漏
fun releaseCodec2() {
    Codec2.destroy(c2instance)
}

额外注意事项 #

• 资源释放：AudioRecord、AudioTrack使用完毕后必须调用release()，避免占用系统音频资源。
• 数据有效性检查：AudioRecord的read()返回值可能为负（表示读取错误），必须处理这种情况，防止无效数据进入编码流程。
• Native资源管理：Codec2实例基于Native实现，不用时必须调用destroy()释放，防止内存泄漏。

内容的提问来源于stack exchange，提问作者snachmsm