如何将混响效果附加到AudioRecord / PCM数据并将其保存到文件中?

时间:2017-01-26 21:50:28

标签: java android pcm android-audiorecord

我想录制来自麦克风的输入,附加混响效果,并将结果保存到文件中。我的用例是一款应用程序,可让您在录制后唱歌并选择不同的预设混响选项,然后保存您的演奏并将其存储在后端服务器上。我发送给服务器的文件需要应用一个混响效果。

到目前为止,我已经能够使用AudioRecord录制输入,并且我可以为AudioTrack添加混响效果以听取混响效果,但我仍然坚持使用如何使用嵌入的混响效果保存音频。这就是我到目前为止所拥有的:

private void startRecording() {
    final int bufferSize = AudioRecord.getMinBufferSize(SAMPLE_RATE, CHANNEL_CONFIG, AUDIO_FORMAT);
    mRecorder = new AudioRecord(MediaRecorder.AudioSource.VOICE_COMMUNICATION, SAMPLE_RATE, CHANNEL_CONFIG, AUDIO_FORMAT, bufferSize);
    if(NoiseSuppressor.isAvailable()) {
        NoiseSuppressor ns = NoiseSuppressor.create(mRecorder.getAudioSessionId());
        ns.setEnabled(true);
    }

    if(AcousticEchoCanceler.isAvailable()) {
        AcousticEchoCanceler aec = AcousticEchoCanceler.create(mRecorder.getAudioSessionId());
        aec.setEnabled(true);
    }

    mPlayer = new AudioTrack(AudioManager.STREAM_MUSIC, SAMPLE_RATE, AudioFormat.CHANNEL_OUT_MONO, AUDIO_FORMAT, bufferSize, AudioTrack.MODE_STREAM);
    PresetReverb reverb = new PresetReverb(1, mPlayer.getAudioSessionId());
    reverb.setPreset(PresetReverb.PRESET_LARGEHALL);
    reverb.setEnabled(true);

    mPlayer.setAuxEffectSendLevel(1.f);

    mRecorder.startRecording();
    mPlayer.play();

    new RecordingThread(mRecorder, mPlayer, bufferSize) {

        @Override
        public void onReleased() {
            mRecorder = null;
            mPlayer = null;
        }
    }.start();
}

public class RecordingThread extends Thread {

    private static final String TAG = RecordingThread.class.getSimpleName();

    private final AudioRecord mRecorder;
    private final AudioTrack mPlayer;
    private final int mBufferSize;

    public RecordingThread(AudioRecord recorder, AudioTrack player, int bufferSize) {
        mRecorder = recorder;
        mPlayer = player;
        mBufferSize = bufferSize;
    }

    @Override
    public void run() {
        byte[] buffer = new byte[mBufferSize];

        int read;
        while ((read = mRecorder.read(buffer, 0, buffer.length)) > 0) {
            mPlayer.write(buffer, 0, buffer.length);
        }

        mRecorder.release();
        mPlayer.release();

        onReleased();

        Timber.tag(TAG);
        Timber.i("Exited with code %s", read);
    }

    public void onReleased() {
    }
}

理想情况下,能够将PresetReverb附加到AudioRecord而不是AudioTrack上会很高兴,但是当我这样做时,我会得到:

java.lang.RuntimeException: Cannot initialize effect engine for type: 47382d60-ddd8-11db-bf3a-0002a5d5c51b Error: -3

所以现在我认为我需要将PCM数据从AudioRecord传输到某个外部服务/库,或者使用任何混响算法来修改缓冲区。

1 个答案:

答案 0 :(得分:5)

根据https://developer.android.com/reference/android/media/audiofx/PresetReverb.html

PresetReverb是一个输出混合辅助效果,应该在音频会话0上创建。为了将MediaPlayer或AudioTrack送入此效果,它们必须明确地附加到它,并且发送级别必须是指定。使用getId()方法返回的效果ID在将其附加到MediaPlayer或AudioTrack时指定此特定效果。

由于此设计用于audiotrack,因此无法将其附加到AudioRecord。我想你有类似的结论。

我建议您使用此SDK https://github.com/superpoweredSDK/Low-Latency-Android-Audio-iOS-Audio-Engine。它支持android和后续效果:

效果:回声,镶边,门,混响,rool,飞快移动,3频段均衡器,双二阶IIR滤波器(低通,高通,带通,高架,低-shelf,参数,缺口)

您还可以考虑https://github.com/ClearVolume/ClearAudio。您可以使用此文件应用混响效果https://github.com/ClearVolume/ClearAudio/blob/master/src/java/clearaudio/synthesizer/filters/ReverbFilter.java

一切顺利!

使用superpoweredSDK添加更多有关如何完成此操作的详细信息。超级SDK有一个名为SuperpoweredReverb的过滤器。其工作方式如下:

enter image description here

让我们通过一些代码看到这一点:

reverb = new SuperpoweredReverb(samplerate);
reverb->enable(true);

reverb->setMix(floatValue); //  Limited to >= 0.0f and <= 1.0f.
reverb->setRoomSize(floatValue); // Limited to >= 0.0f and <= 1.0f.

SuperpoweredShortIntToFloat(intBuffer, floatBuffer, numberOfSamples);

bool res = reverb->process(floatBuffer, floatBuffer, numberOfSamples);

SuperpoweredFloatToShortInt(floatBuffer, intBuffer, numberOfSamples);

让我尽力解释这里发生的事情:

创建过滤器

reverb = new SuperpoweredReverb(samplerate);
reverb->enable(true);

可选:对此过滤器应用其他选项,例如

reverb->setMix(floatValue); //  Limited to >= 0.0f and <= 1.0f.
reverb->setRoomSize(floatValue); // Limited to >= 0.0f and <= 1.0f.

获取源样本;如果样本是整数,请使用SuperpoweredShortIntToFloat进行转换。这里intBuffer有输入音频样本,numberOfSamples是样本数。 floatBuffer将包含浮点转换缓冲区

SuperpoweredShortIntToFloat(intBuffer, floatBuffer, numberOfSamples);

调用过程以应用混响效果。此方法将应用混响效果并将输出再次复制到floatBuffer

bool res = reverb->process(floatBuffer, floatBuffer, numberOfSamples);

要将其转换回Int,请使用以下方法

SuperpoweredFloatToShortInt(floatBuffer, intBuffer, numberOfSamples);

然后你可以将这个混响应用的音频样本存储在你想要的任何地方,无论是在文件中还是播放

来自superpoweredSDK的更多发现,它还具有录制功能。见http://superpowered.com/docs/class_superpowered_recorder.html

一个例子:

recorder = new SuperpoweredRecorder(temporaryPath, samplerate);
recorder->start(destinationPath);

// With input samples
SuperpoweredShortIntToFloat(intBuffer, floatBuffer, numberOfSamples);

bool res = recorder->process(floatBuffer, NULL, numberOfSamples);

SuperpoweredFloatToShortInt(floatBuffer, intBuffer, numberOfSamples);