how can we enhance audio quality by adding pitch and pause factor with intensity in spectrogram

[anil]

protected void computeSpectrogram() {
try {
AudioDataInputStream is = new AudioDataInputStream(audio);
dataSource.setInputStream(is, "live audio");

        /* Run through all the spectra one at a time and convert
         * them to an log intensity value.
         */
        ArrayList<double[]> intensitiesList = new ArrayList<double[]>();  //intensity List
        double maxIntensity = Double.MIN_VALUE;
        Data spectrum = frontEnd.getData(); //get data

        while (!(spectrum instanceof DataEndSignal)) {
            if (spectrum instanceof DoubleData) {
                double[] spectrumData = ((DoubleData) spectrum).getValues();
             //   System.out.println("values................."+spectrumData);
                double[] intensities = new double[spectrumData.length];
                for (int i = 0; i < intensities.length; i++) {
                    /*
         * A very small intensity is, for all intents
                     * and purposes, the same as 0.
                     */
                    intensities[i] = Math.max(Math.log(spectrumData[i]),
                            0.0);
                    if (intensities[i] > maxIntensity) {
                        maxIntensity = intensities[i];
                    }
                }
                intensitiesList.add(intensities);
            }
            spectrum = frontEnd.getData();
        }
        is.close();

        int width = intensitiesList.size();
        int height = (intensitiesList.get(0)).length;
        int maxYIndex = height - 1;
        Dimension d = new Dimension(width, height);

        setMinimumSize(d);
        setMaximumSize(d);
        setPreferredSize(d);

        /* Create the image for displaying the data.
         */
        spectrogram = new BufferedImage(width,
                height,
                BufferedImage.TYPE_INT_RGB);

        /* Set scaleFactor so that the maximum value, after removing
        * the offset, will be 0xff.
        */
        double scaleFactor = ((0xff + offsetFactor) / maxIntensity);

        for (int i = 0; i < width; i++) {
            double[] intensities = intensitiesList.get(i);
            for (int j = maxYIndex; j >= 0; j--) {

                /* Adjust the grey value to make a value of 0 to mean
                * white and a value of 0xff to mean black.
                */
                int grey = (int) (intensities[j] * scaleFactor
                        - offsetFactor);
                grey = Math.max(grey, 0);
                grey = 0xff - grey;

                /* Turn the grey into a pixel value.
                */
                int pixel = ((grey << 16) & 0xff0000)
                        | ((grey << 8) & 0xff00)
                        | (grey & 0xff);

                spectrogram.setRGB(i, maxYIndex - j, pixel);
            }
        }
        ImageFilter scaleFilter =
                new ReplicateScaleFilter((int) (zoom * width), height);
        scaledSpectrogram =
                createImage(new FilteredImageSource(spectrogram.getSource(),
                        scaleFilter));
        Dimension sz = getSize();
        repaint(0, 0, 0, sz.width - 1, sz.height - 1);
    } catch (Exception e) {
        e.printStackTrace();
    }
}

[Nickolay V. Shmyrev]

What your question is about?

[anil]

i mean if user speaks too low or too loud ...can the recognizer be able to recognize,....what are its parameters