Well, the only thought that comes to my mind is that 1, 2, and 3 use the lowest frequency of 697 Hz. It might be that due to the small speakers this frequency is played back at a significantly lower volume (This difference must be quite big, because DTMF decoders should be robust against this).

In order to fix this I would have to implement frequency dependent amplitudes.

Anyone with a good measurement set-up to check if this is the reason for this problem? I would need some numbers on what the amplitudes should be ...