Voice-activated selective memory for voice-capturing devices

A computer-implemented method, comprising:

• starting a private session based at least in part on input from one of one or more voice-capturing devices, wherein the private session comprises a session configured to automatically delete representations of received inputs based on an end of the private session;
• receiving, at a service provider environment comprising one or more services, a voice input from at least one of the one or more voice-capturing devices via a network, wherein the voice input is received while in the private session, and wherein the voice input comprises one or more utterances from a user of the one or more voice-capturing devices;
• storing and analyzing a representation of the voice input in the service provider environment;
• and
• after storing and analyzing the representation of the voice input and based on the end of the private session, automatically deleting, without being directly prompted by user input, the stored representation of the voice input based at least in part on the voice input having been received while in the private session.

One or more non-transitory computer-readable storage media storing program instructions that, when executed on or across one or more processors, perform:

• starting a private session based at least in part on input from one of one or more voice-capturing devices, wherein the private session comprises a session configured to automatically delete representations of received inputs based on an end of the private session;
• receiving, at a service provider environment comprising one or more services, a voice input from at least one of the one or more voice-capturing devices via a network, wherein the voice input is received while in the private session, and wherein the voice input comprises one or more utterances from a user of the one or more voice-capturing devices;
• storing and analyzing a representation of the voice input in the service provider environment;
• and
• after storing and analyzing the representation of the voice input and based on the end of the private session, automatically deleting, without being directly prompted by user input, the stored representation of the voice input based at least in part on the voice input having been received while in private session.