Enhancing Speech-to-Speech Translation with Multiple TTS Targets