在Actionscript中,音频录制文件已损坏

Question

我使用Adobe Flash Builder 4.6/AIR从麦克风录制了语音样本,成功录制了语音.我首先在actionscript中将语音数据(字节数组)转换为base64格式,然后使用我的PHP代码将该base64数据转换为WAV文件.但是在RiffPad中WAV文件抛出文件已损坏.

RIFFPad是RIFF格式文件的查看器,如WAV,AVI.

预期的wav文件规范:

采样率:22KHZ

    // -- saves the current audio data as a .wav file
    protected function onSubmit( event:Event ):void {
        alertBox.show("Processing ... please wait.");

        stopPlayback();
        stopRecording();
        playBtn.enabled = recordBtn.enabled = submitBtn.enabled = false;
        var position:int = capture.buffer.position;
        var wavWriter:WAVWriter = new WAVWriter()
        var wavWriter1:WaveEncoder = new WaveEncoder()
        wavWriter.numOfChannels = 1;
        wavWriter.samplingRate = 22050;
        wavWriter.sampleBitRate = 16; 
        var wavBytes:ByteArray = new ByteArray;
        capture.buffer.position = 0;
        wavWriter.processSamples(wavBytes, capture.buffer, capture.microphone.rate * 1000, 1);
        Settings.alertBox3.show("RATE :"+capture.microphone.rate); //Here show RATE: 8
        //wavWriter.processSamples(wavBytes, capture.buffer, 22050, 1);
        //wavBytes = wavWriter1.encode( capture.buffer, 1, 16, 22050);
        capture.buffer.position = position;
        wavBytes.position=0;
        submitVoiceSample(Base64_new.encodeByteArray(wavBytes));
    }

WAV Writer头功能:

public var samplingRate = 22050;
public var sampleBitRate:int = 8;
public var numOfChannels:int = 2;
private var compressionCode:int = 1;

private function header(dataOutput:IDataOutput, fileSize:Number):void
{
    dataOutput.writeUTFBytes("RIFF");
    dataOutput.writeUnsignedInt(fileSize); // Size of whole file
    dataOutput.writeUTFBytes("WAVE");
    // WAVE Chunk
    dataOutput.writeUTFBytes("fmt ");   // Chunk ID
    dataOutput.writeUnsignedInt(16);    // Header Chunk Data Size
    dataOutput.writeShort(compressionCode); // Compression code - 1 = PCM
    dataOutput.writeShort(numOfChannels); // Number of channels
    dataOutput.writeUnsignedInt(samplingRate); // Sample rate
    dataOutput.writeUnsignedInt(samplingRate * numOfChannels * sampleBitRate / 8); // Byte Rate == SampleRate * NumChannels * BitsPerSample/8       
    dataOutput.writeShort(numOfChannels * sampleBitRate / 8); // Block align == NumChannels * BitsPerSample/8
    dataOutput.writeShort(sampleBitRate); // Bits Per Sample
}

WAV文件编写器功能:

public function processSamples(dataOutput:IDataOutput, dataInput:ByteArray, inputSamplingRate:int, inputNumChannels:int = 1):void
{
    if (!dataInput || dataInput.bytesAvailable <= 0) // Return if null
        throw new Error("No audio data");

    // 16 bit values are between -32768 to 32767.
    var bitResolution:Number = (Math.pow(2, sampleBitRate)/2)-1;
    var soundRate:Number = samplingRate / inputSamplingRate;
    var dataByteLength:int = ((dataInput.length/4) * soundRate * sampleBitRate/8);
    // data.length is in 4 bytes per float, where we want samples * sampleBitRate/8 for bytes
    //var fileSize:int = 32 + 8 + dataByteLength;
    var fileSize:int = 32 + 4 + dataByteLength;
    // WAV format requires little-endian
    dataOutput.endian = Endian.LITTLE_ENDIAN;  
    // RIFF WAVE Header Information
    header(dataOutput, fileSize);
    // Data Chunk Header
    dataOutput.writeUTFBytes("data");
    dataOutput.writeUnsignedInt(dataByteLength); // Size of whole file

    // Write data to file
    dataInput.position = 0;
    var tempData:ByteArray = new ByteArray();
    tempData.endian = Endian.LITTLE_ENDIAN;

    // Write to file in chunks of converted data.
    while (dataInput.bytesAvailable > 0) 
    {
        tempData.clear();
        // Resampling logic variables
        var minSamples:int = Math.min(dataInput.bytesAvailable/4, 8192);
        var readSampleLength:int = minSamples;//Math.floor(minSamples/soundRate);
        var resampleFrequency:int = 100;  // Every X frames drop or add frames
        var resampleFrequencyCheck:int = (soundRate-Math.floor(soundRate))*resampleFrequency;
        var soundRateCeil:int = Math.ceil(soundRate);
        var soundRateFloor:int = Math.floor(soundRate);
        var jlen:int = 0;
        var channelCount:int = (numOfChannels-inputNumChannels);
        /*
        trace("resampleFrequency: " + resampleFrequency + " resampleFrequencyCheck: " + resampleFrequencyCheck
            + " soundRateCeil: " + soundRateCeil + " soundRateFloor: " + soundRateFloor);
        */
        var value:Number = 0;
        // Assumes data is in samples of float value
        for (var i:int = 0;i < readSampleLength;i+=4)
        {
            value = dataInput.readFloat();
            // Check for sanity of float value
            if (value > 1 || value < -1)
                throw new Error("Audio samples not in float format");

            // Special case with 8bit WAV files
            if (sampleBitRate == 8)
                value = (bitResolution * value) + bitResolution;
            else
                value = bitResolution * value;

            // Resampling Logic for non-integer sampling rate conversions
            jlen = (resampleFrequencyCheck > 0 && i % resampleFrequency < resampleFrequencyCheck) ? soundRateCeil : soundRateFloor; 
            for (var j:int = 0; j < jlen; j++)
            {
                writeCorrectBits(tempData, value, channelCount);
            }
        }
        dataOutput.writeBytes(tempData);
    }
}

我将base64数据发送到我的服务请求 php侧我得到'$ this-> request-> voiceSample'参数并解码base64到.wav文件

 file_put_contents('name.wav', base64_decode($this->request->voiceSample));

在Riffpad中加载"name.wav"文件后 我遇到了问题

文件末尾有额外的垃圾.

请任何人给我解决这个问题的建议......

Answer 1

这行代码有一个固有的错误：

 wavWriter.processSamples(wavBytes, capture.buffer, capture.microphone.rate * 1000, 1);

该Microphone.rate手册指出实际采样频率与microphone.rate*1000此代码预期的不同。实际表如下：

rate   Actual frequency
44     44,100 Hz
22     22,050 Hz
11     11,025 Hz
8      8,000 Hz
5      5,512 Hz

因此，虽然您的代码注释指出报告rate为 8，但一般情况下客户端的情况可能并非如此，因此请在将推导的采样率传递到wavWriter.processSamples().

接下来，您dataByteLength通过浮点计算进行预先计算，当您逐字节采样数据时，这可能最终不准确，因此最好首先重新采样，然后收集数据长度，然后将所有数据写入dataOutput，如下所示：

public function processSamples(dataOutput:IDataOutput, dataInput:ByteArray, inputSamplingRate:int, inputNumChannels:int = 1):void
{
    if (!dataInput || dataInput.bytesAvailable <= 0) // Return if null
        throw new Error("No audio data");

    // 16 bit values are between -32768 to 32767.
    var bitResolution:Number = (Math.pow(2, sampleBitRate)/2)-1;
    // var soundRate:Number = samplingRate / inputSamplingRate;
    // var fileSize:int = 32 + 4 + dataByteLength; kept for reference
    // fmt tag is 4+4+16, data header is 8 bytes in size, and 4 bytes for WAVE
    // but the data length is not yet determined
    // WAV format requires little-endian
    dataOutput.endian = Endian.LITTLE_ENDIAN;  
    // Prepare data for data to file
    dataInput.position = 0;
    var tempData:ByteArray = new ByteArray();
    tempData.endian = Endian.LITTLE_ENDIAN;
    // Writing in chunks is no longer possible, because we don't have the header ready

    // Let's precalculate the data needed in the loop
    var step:Number=inputSamplingRate / samplingRate; // how far we should step into the input data to get next sample
    var totalOffset:Number=1.0-1e-8; // accumulator for step
    var oldChannels:Array=[];
    var i:int;
    for (i=0;i<numOfChannels;i++) oldChannels.push(0.0);
    // previous channels' sample holder
    var newChannels:Array=oldChannels.slice(); // same for new channels that are to be read from byte array
    // reading first sample set from input byte array
    if (dataInput.bytesAvailable>=inputNumChannels*4) {
        for (i=0;i<inputNumChannels;i++) {
            var buf:Number=dataInput.readFloat();
            if (buf > 1) buf=1; if (buf < -1) buf=-1;
            newChannels[i]=buf;
        }
        // if there's one channel, copy data to other channels
        if ((inputNumChannels==1) && (numOfChannels>1)) {
            for (i=1;i<numOfChannels;i++) newChannels[i]=newChannels[0];                
        }
    }
    while ((dataInput.bytesAvailable>=inputNumChannels*4) || (totalOffset<1.0))
    {
        // sample next value for output wave file
        var value:Number;
        for (i=0;i<numOfChannels;i++) {
            value = (totalOffset*newChannels[i])+(1.0-totalOffset)*oldChannels[i];
            // linear interpolation between old sample and new sample
            // Special case with 8bit WAV files
            if (sampleBitRate == 8)
                value = (bitResolution * value) + bitResolution;
            else
                value = bitResolution * value; 
            // writing one channel into tempData
            writeCorrectBits(tempData, value, 0);
        }
        totalOffset+=step; // advance per output sample
        while ((totalOffset>1) && (dataInput.bytesAvailable>=inputNumChannels*4)) {
            // we need a new sample, and have a sample to process in input
            totalOffset-=1;
            for (i=0;i<numOfChannels;i++) oldChannels[i]=newChannels[i]; // store old sample
            // get another sample, copypasted from above
            for (i=0;i<inputNumChannels;i++) {
                value=dataInput.readFloat();
                if (value > 1) value=1; if (value < -1) value=-1; // sanity check
                // I made it clip instead of throwing exception, replace if necessary
                // if (value > 1 || value < -1) throw new Error("Audio samples not in float format");
                newChannels[i]=value;
            }
            if ((inputNumChannels==1) && (numOfChannels>1)) {
                for (i=1;i<numOfChannels;i++) newChannels[i]=newChannels[0];
            }
        } // end advance by totalOffset
    } // end main loop
    var dataBytesLength:uint=tempData.length; // now the length will be correct by definition
    header(dataOutput, 32+4+dataBytesLength);
    dataOutput.writeUTFBytes("data");
    dataOutput.writeUnsignedInt(dataBytesLength);
    dataOutput.writeBytes(tempData);

}

我已经重写了重新采样例程以使用滑动窗口算法（如果新采样率高于旧采样率，但接受任何比率，效果最好）。该算法在样本之间使用线性插值，而不是简单地在插值序列的长度上重复使用旧样本。请随意替换为您自己的循环。应该保留的原则是，首先编译完整的 tempData，然后才用现在正确定义的数据长度写入标头。

如有问题请报告。