How is the spoken output for VoiceXML applications stored?

Pre-recorded audio files are stored as headerless μ-law, headerless a-law, .wav, or .au files on the Web or application server. Audio files referenced in a VoiceXML document are automatically downloaded and cached by the VoiceXML browser until needed. Cached audio files are automatically deleted from the cache after an expiry time.

For more information about text-to-speech, see Text-to-speech