Allows the recognition of user voice input through the browser.

Voice recognition must be used in conjunction with long polling (see <long_poll connect="">) in order to capture the voice input and evaluate the result.

Note that the <next_node> specified in <voice_rec> is the node during which the actual recognition will take place. Therefore, the <timeout> specified in <voice_rec> will occur relative to the start of the <next_node>, not the node in which <voice_rec> is defined. However, if <mic_delayed> is set to YES, the recognition and timeout will not begin until the recognition is explicitly started using <voice_rec_start/>.

Once a <voice_rec> tag has been used, the variable max_level is available. max_level is the highest volume out of 100 that the user's voice input reached. Input that scores too high may result in distortion; input that scores too low may be too soft to be recognized. Note that if you try to use max_level before using <voice_rec>, an unknown variable error will occur.

Child of: <click>, <mouse_click>, <mouse_down>, <mouse_move>, <mouse_out>, <mouse_over>, <node name="">, <press>, <release>
Parent of: <timeout>, <server>, <grammar>, <dynamic_grammar>, <next_node>, <error_node>, <all_full_node>, <mic_deny_node>, <mic_delayed>, <recording_start start_level="15">
See also: <long_poll connect="">


            <grammar>yesorno, largeorsmall</grammar>
            <dynamic_grammar>PTN_1:RET_1, PTN_2:RET_2</dynamic_grammar>
            <recording_start start_level="15">

To stop a voice recognition session independently of <timeout>:


To stop a voice recognition session and close the connection without receiving a return value from the long polling session, use CANCEL as follows: