Embodiments include methods and apparatus for synchronizing data and focus between visual and voice views associated with distributed multi-modal applications. An embodiment includes a client device adapted to render a visual display that includes at least one multi-modal display element for which input data is receivable though a visual modality and a voice modality. When the client detects a user utterance via the voice modality, the client sends uplink audio data representing the utterance to a speech recognizer. An application server receives a speech recognition result generated by the speech recognizer, and sends a voice event response to the client. The voice event response is sent as a response to an asynchronous HTTP voice event request previously sent to the application server by the client. The client may then send another voice event request to the application server in response to receiving the voice event response.