Systems and methods for voice synthesis are disclosed for providing a synthesized voice message that is consonant with the taste of a customer and a program storage device readable by machine to perform method steps for voice synthesis. In accordance with an order from a customer received via a network, a service provider generates voice synthesis data, based on voice characteristic data for a speaker chosen by the customer, that is produced for a sentence input by the customer, and prepares to deliver the voice synthesis data to the customer. At this time, a transaction number is provided for the order received from the customer, and subsequently, when the transaction number is presented by the customer, the generated voice synthesis data are delivered to the customer. The customer then loads the received voice synthesis data into a device that reproduces the voiced sentence.