Generating digital avatar

An apparatus comprising:

• one or more non-transitory computer-readable storage media including instructions;
• and
• one or more processors coupled to the storage media, the one or more processors configured to execute the instructions to:
• receive one or more non-video inputs, wherein the one or more non-video inputs comprises at least one of a text input, an audio input, or an expression input;
• access a K-NN graph comprising a plurality of sets of nodes, wherein each set of nodes corresponds to a particular semantic context of a plurality of semantic contexts;
• process the one or more non-video inputs using the K-NN graph to identify one or more semantic contexts corresponding to one or more sets of nodes, respectively, that relate to the one or more non-video inputs;
• determine one or more actions to be performed by a digital avatar based on the one or more identified semantic contexts;
• generate, in real-time responsive to receiving the one or more non-video inputs and based on the determined one or more actions, a video output of the digital avatar comprising one or more human characteristics corresponding to the one or more identified semantic contexts;
• and
• send, to a client device, instructions to present the video output of the digital avatar.

A computer-readable non-transitory storage media comprising instructions executable by a processor to:

• receive one or more non-video inputs, wherein the one or more non-video inputs comprises at least one of a text input, an audio input, or an expression input;
• access a K-NN graph comprising a plurality of sets of nodes, wherein each set of nodes corresponds to a particular semantic context of a plurality of semantic contexts;
• process the one or more non-video inputs using the K-NN graph to identify one or more semantic contexts corresponding to one or more sets of nodes, respectively, that relate to the one or more non-video inputs;
• determine one or more actions to be performed by a digital avatar based on the one or more identified semantic contexts;
• generate, in real-time responsive to receiving the one or more non-video inputs and based on the determined one or more actions, a video output of the digital avatar comprising one or more human characteristics corresponding to the one or more identified semantic contexts;
• and
• send, to a client device, instructions to present the video output of the digital avatar.