Efficient Multispeaker Speech Synthesis And Voice Cloning