Increase volume output of recorded audio

I am trying to make a call recording app in Android. I am using loudspeaker to record both uplink and downlink audio. The only problem I am facing is the volume is too low. I've increased the volume of device using AudioManager to max and it can't go beyond that.

I've first used MediaRecorder, but since it had limited functions and provides compressed audio, I've tried with AudioRecorder. Still I havn't figured out how to increase the audio. I've checked on projects on Github too, but it's of no use. I've searched on stackoverflow for last two weeks, but couldn't find anything at all.

I am quite sure that it's possible, since many other apps are doing it. For instance Automatic Call recorder does that.

I understand that I have to do something with the audio buffer, but I am not quite sure what needs to be done on that. Can you guide me on that.

Update:-
I am sorry that I forgot to mention that I am already using Gain. My code is almost similar to RehearsalAssistant (in fact I derived it from there). The gain doesn't work for more than 10dB and that doesn't increase the audio volume too much. What I wanted is I should be able to listen to the audio without putting my ear on the speaker which is what lacking in my code.

I've asked a similar question on functioning of the volume/loudness at SoundDesign SE here. It mentions that the Gain and loudness is related but it doesn't set the actual loudness level. I am not sure how things work, but I am determined to get the loud volume output.

You obviously have the AudioRecord stuff running, so I skip the decision for sampleRate and inputSource. The main point is that you need to appropriately manipulate each sample of your recorded data in your recording loop to increase the volume. Like so:

    int minRecBufBytes = AudioRecord.getMinBufferSize( sampleRate, AudioFormat.CHANNEL_IN_MONO, AudioFormat.ENCODING_PCM_16BIT );
    // ...
    audioRecord = new AudioRecord( inputSource, sampleRate, AudioFormat.CHANNEL_IN_MONO, AudioFormat.ENCODING_PCM_16BIT, minRecBufBytes );

    // Setup the recording buffer, size, and pointer (in this case quadruple buffering)
    int recBufferByteSize = minRecBufBytes*2;
    byte[] recBuffer = new byte[recBufferByteSize];
    int frameByteSize = minRecBufBytes/2;
    int sampleBytes = frameByteSize;
    int recBufferBytePtr = 0;

    audioRecord.startRecording();

    // Do the following in the loop you prefer, e.g.
    while ( continueRecording ) {
        int reallySampledBytes = audioRecord.read( recBuffer, recBufferBytePtr, sampleBytes );

        int i = 0;
        while ( i < reallySampledBytes ) {
            float sample = (float)( recBuffer[recBufferBytePtr+i  ] & 0xFF
                                  | recBuffer[recBufferBytePtr+i+1] << 8 );

            // THIS is the point were the work is done:
            // Increase level by about 6dB:
            sample *= 2;
            // Or increase level by 20dB:
            // sample *= 10;
            // Or if you prefer any dB value, then calculate the gain factor outside the loop
            // float gainFactor = (float)Math.pow( 10., dB / 20. );    // dB to gain factor
            // sample *= gainFactor;

            // Avoid 16-bit-integer overflow when writing back the manipulated data:
            if ( sample >= 32767f ) {
                recBuffer[recBufferBytePtr+i  ] = (byte)0xFF;
                recBuffer[recBufferBytePtr+i+1] =       0x7F;
            } else if ( sample <= -32768f ) {
                recBuffer[recBufferBytePtr+i  ] =       0x00;
                recBuffer[recBufferBytePtr+i+1] = (byte)0x80;
            } else {
                int s = (int)( 0.5f + sample );  // Here, dithering would be more appropriate
                recBuffer[recBufferBytePtr+i  ] = (byte)(s & 0xFF);
                recBuffer[recBufferBytePtr+i+1] = (byte)(s >> 8 & 0xFF);
            }
            i += 2;
        }

        // Do other stuff like saving the part of buffer to a file
        // if ( reallySampledBytes > 0 ) { ... save recBuffer+recBufferBytePtr, length: reallySampledBytes

        // Then move the recording pointer to the next position in the recording buffer
        recBufferBytePtr += reallySampledBytes;

        // Wrap around at the end of the recording buffer, e.g. like so:
        if ( recBufferBytePtr >= recBufferByteSize ) {
            recBufferBytePtr = 0;
            sampleBytes = frameByteSize;
        } else {
            sampleBytes = recBufferByteSize - recBufferBytePtr;
            if ( sampleBytes > frameByteSize )
                sampleBytes = frameByteSize;
        }
    }