Curating audio and IR commands through machine learning

Techniques for curating audio and IR commands using machine learning may be provided. For example, the system can receive an audio stream that includes a plurality of audio segments and the system can store the audio stream and/or segments. The system can also store a command in a second data store. When a portion of the audio is provided in conjunction with a same command around the same time and exceeds a threshold number of repetitions, the next time that that audio segment is received, the system may provide a command that corresponds with that audio segment to an output device to cause an operation of the output device. In some examples, the system may confirm that the audio segment corresponds with the command before providing the command. This disclosure may use, for example, signal detection, acoustic fingerprinting, and shared vocabulary lists.