Multi-modal interface in a voice-activated network

A method of controlling digital components in voice-activated environments, comprising:

• receiving a first input audio signal acquired via a sensor of a client device;
• parsing the first input audio signal to identify a request for content;
• identifying, from a plurality of digital components, a digital component based on the request for content identified from the first input audio signal, the digital component having a plurality of portions defined by a plurality of break points generated based on a set of annotations;
• receiving, by a data processing system, a second input audio signal acquired via the sensor of the client device, wherein the second input audio signal comprises a natural language input from a user;
• parsing, by the data processing system, the second input audio signal to identify a keyword from the second input audio signal;
• identifying, by the data processing system, from the set of annotations, an annotation associated with the keyword identified from the second input audio signal;
• selecting, by the data processing system, from the plurality of portions of the digital component, a portion corresponding to the annotation that is identified as associated with the keyword;
• and
• providing, by the data processing system, the portion selected from the plurality of portions to the client device.