A system may include first and second speech-processing systems with corresponding first and second wakewords. An utterance may contain two or more wakewords. The system determines which wakeword was spoken first and can send data to that wakeword's speech-processing system to perform further processing.