Java - Adjust playback speed of a WAV file

Question

I'm likely dense but I cannot seem to find a solution to my issue

(NOTE: I CAN find lots of people reporting this issue, seems like it happened as a result of newer Java (possible 1.5?). Perhaps SAMPLE_RATE is no longer supported? I am unable to find any solution).

I'm trying to adjust the SAMPLE_RATE to speed up/slow down song. I can successfully play a .wav file without issue, so I looked into FloatControl which worked for adjusting volume:

public void adjustVolume(String audioType, float gain) {
        FloatControl gainControl = null;

        gainControl = (FloatControl) clipSFX.getControl(FloatControl.Type.MASTER_GAIN);
                if(gain > MAX_VOLUME)
                    gain = MAX_VOLUME;
                if(gain < MIN_VOLUME)
                    gain = MIN_VOLUME;

            //set volume
            gainControl.setValue(gain);
    }

But when trying to translate this principle to SAMPLE_RATE, I get an error very early on at this stage:

    public void adjustVolume(String audioType, float gain) {
        FloatControl gainControl = null;

        gainControl = (FloatControl) clipSFX.getControl(FloatControl.Type.SAMPLE_RATE);
        //ERROR: Exception in thread "Thread-3" java.lang.IllegalArgumentException: Unsupported control type: Sample Rate

        //I haven't gotten this far yet since the above breaks, but in theory will then set value?
            gainControl.setValue(gain);
}

Everything I've found online seems to be related to taking input from a mic or some external line and doesn't seem to translate to using an audio file, so I'm unsure what I'm missing. Any help would be appreciated! Thanks!

Answer 1

Here we have a method that changes the speed - by doubling the sample rate. Basically the steps are as follows:

• open the audio stream of the file
• get the format
• create a new format with the sample rate changed
• open a data line with that format
• read from the file/audio stream and play onto the line

The concepts here are SourceDataLine, AudioFormat and AudioInputStream. If you look at the javax.sound tutorial you will find them, or even the pages of the classes. You can now create your own method (like adjust(factor)) that just gets the new format and all else stay the same.

  public void play() {
    try {
      File fileIn = new File(" ....);
      AudioInputStream audioInputStream=AudioSystem.getAudioInputStream(fileIn);
      AudioFormat formatIn=audioInputStream.getFormat();
      AudioFormat format=new AudioFormat(formatIn.getSampleRate()*2, formatIn.getSampleSizeInBits(), formatIn.getChannels(), true, formatIn.isBigEndian());
          System.out.println(formatIn.toString());
          System.out.println(format.toString());
      byte[] data=new byte[1024];
      DataLine.Info dinfo=new DataLine.Info(SourceDataLine.class, format);
      SourceDataLine line=(SourceDataLine)AudioSystem.getLine(dinfo);
      if(line!=null) {
        line.open(format);
        line.start();
        while(true) {
          int k=audioInputStream.read(data, 0, data.length);
          if(k<0) break;
          line.write(data, 0, k);
        }
        line.stop();
        line.close();
      }
    }
    catch(Exception ex) { ex.printStackTrace(); }
  }

Answer 2

It is also possible to vary the speed by using linear interpolation when progressing through the audio data.

Audio values are laid out in an array and the cursor normally goes from value to value. But you can set things up to progress an arbitrary amount, for example 1.5 frames, and create a weighted value where needed.

Suppose data is as follows:

1. 0.5
2. 0.8
3. 0.2
4. -0.1
5. -0.5
6. -0.7

Your playback data (for 1.5 rate) would be

1. 0.5
2. (0.8 + 0.2)/2
3. -0.1
4. (-0.5 + -0.7)/2

I know there have been posts that more fully explain this algorithm before on Stack Overflow. Forgive me for not tracking them down.

I use this method to allow real-time speed changes in .wav playback in the following open-source library: AudioCue. Feel free to check out the code and make use of the ideas in it.

Following is the method that creates a stereo pair of audio values from a spot that lies in between two audio frames (data is signed floats, ranging from -1 to 1). It's from an inner class AudioCuePlayer in AudioCue.java. Probably not the easiest to read. The sound data being read is in the array cue, and idx is the current "play head" location that is progressing through this array. 'intIndex' is the audio frame, and 'flatIndex' is the actual location of the frame in the array. I use frames to track the playhead's location and calculate the interpolation weights, and then use the flatIndex for getting the corresponding values from the array.

private float[] readFractionalFrame(float[] audioVals, float idx)
{
    final int intIndex = (int) idx;
    final int flatIndex = intIndex * 2;

    audioVals[0] = cue[flatIndex + 2] * (idx - intIndex) 
            + cue[flatIndex] * ((intIndex + 1) - idx);

    audioVals[1] = cue[flatIndex + 3] * (idx - intIndex) 
            + cue[flatIndex + 1] * ((intIndex + 1) - idx);

    return audioVals;
}

I'd be happy to clarify if there are questions.