The invention discloses a method for enhancing the performance of playing remote voice by voice extensible sign language business, which is applied to a phone voice system. The method comprises the following steps: an agent module is arranged on a VXML platform in the phone voice system; when a service needs to play a voice document, a VXML analyzer sends a request to the agent module; the agent module firstly queries that whether the voice document is downloaded or not after receiving the request of the VXML analyzer, and returns the voice document to the VXML analyzer when the voice document is downloaded; the agent module requests a remoter server for downloading the voice document when the voice document is not downloaded, returns the voice document to the VXML analyzer when the download is completed, and buffers the information of the voice document at the same time. The invention introduces the agent module to improve the remote voice performance of a VXML business private room by utilizing buffer mechanism, reduce system burden and solve the problem of waste of system resources.