如何使用TarsosDSP获得MFCC?

时间:2016-11-05 08:06:38

标签: android mfcc tarsosdsp

我到处搜索,我无法弄清楚如何在Android上使用TarsosDSP提取MFCC功能。我知道如何从文件中获取FFT。 有什么帮助吗?

2 个答案:

答案 0 :(得分:1)

见官方github page

MFCC测试文件

public class MFCCTest {

//  private static int counter = 0;

    @Test
    public void MFCCForSineTest() throws UnsupportedAudioFileException{
        int sampleRate = 44100;
        int bufferSize = 1024;
        int bufferOverlap = 128;
        final float[] floatBuffer = TestUtilities.audioBufferSine();
        final AudioDispatcher dispatcher = AudioDispatcherFactory.fromFloatArray(floatBuffer, sampleRate, bufferSize, bufferOverlap);
        final MFCC mfcc = new MFCC(bufferSize, sampleRate, 40, 50, 300, 3000);
        dispatcher.addAudioProcessor(mfcc);
        dispatcher.addAudioProcessor(new AudioProcessor() {

            @Override
            public void processingFinished() {
            }

            @Override
            public boolean process(AudioEvent audioEvent) {
                return true;
            }
        });
        dispatcher.run();
    }

}

和TestUtilities audioBufferSine()

public class TestUtilities {

    /**
     * Constructs and returns a buffer of a two seconds long pure sine of 440Hz
     * sampled at 44.1kHz.
     * 
     * @return A buffer of a two seconds long pure sine (440Hz) sampled at
     *         44.1kHz.
     */
    public static float[] audioBufferSine() {
        final double sampleRate = 44100.0;
        final double f0 = 440.0;
        final double amplitudeF0 = 0.5;
        final double seconds = 4.0;
        final float[] buffer = new float[(int) (seconds * sampleRate)];
        for (int sample = 0; sample < buffer.length; sample++) {
            final double time = sample / sampleRate;
            buffer[sample] = (float) (amplitudeF0 * Math.sin(2 * Math.PI * f0 * time));
        }
        return buffer;
    }

答案 1 :(得分:0)

你可以在进程事件下获得mfcc,我认为它适用于每一帧

    int sampleRate = 16000;
    int bufferSize = 512;
    int bufferOverlap = 128;
    new AndroidFFMPEGLocator(this);
    final List<float[]>mfccList = new ArrayList<>(200);
    InputStream inStream = new FileInputStream(audioFilePath);
   AudioDispatcher dispatcher = new AudioDispatcher(new UniversalAudioInputStream(inStream, new TarsosDSPAudioFormat(sampleRate, bufferSize, 1, true, true)), bufferSize, bufferOverlap);
    final MFCC mfcc = new MFCC(bufferSize, sampleRate, 20, 50, 300, 3000);
    dispatcher.addAudioProcessor(mfcc);
    dispatcher.addAudioProcessor(new AudioProcessor() {

        @Override
        public void processingFinished() {
        }

        @Override
        public boolean process(AudioEvent audioEvent) {
            mfccList.add( mfcc.getMFCC());
            return true;
        }
    });
    dispatcher.run();