Natural conversation storytelling system

A multi-media interactive story-telling system, said system comprising:

• an interview intake device for a multi-media story-telling system, said interview intake device comprising:
• a display device for displaying questions defined in a script;
• recording devices for recording one or more interview sessions in which one or more contributors answer the questions displayed on the display device;
• wherein the one or more contributors are presented with an initial recorded interview session and to review, trim, filter, or add to the initial recorded interview session;
• processors configured to generate metadata for the one or more recorded interview sessions for cataloguing and retrieval;
• processors to add additional media files and vox-tags to trigger display of the additional media files;
• and
• wherein the one or more recorded interview sessions are automatically transcribed and transcripts are used as part of metadata of the one or more interview sessions;
• and an interactive device comprising one or more processors:
• to compare a query from a user to a plurality of existing queries and to use machine learning to identify similar wording and similar responses;
• to use the machine learning to match the query to retrieve and play the recorded session;
• and
• wherein the interactive device uses a machine learning algorithm to predict a follow-up query that is most likely to follow the query and to retrieve and cache a predicted recorded session corresponding to a response to the follow-up query during the play of the recorded session, wherein the machine learning algorithm is trained on past patterns of inquiries.

A multi-media interactive device, said multi-media interactive device comprising:

• an input device that features a user interface, said input device configured to receive a query from a user;
• an output device for playing back recorded interview sessions, wherein said interview sessions are recorded by one or more contributors and wherein in each interview session the one or more contributors answer a question defined in a script;
• processors configured to retrieve a recorded interview session in response to the query, wherein in the recorded interview session the answer from the one or more contributors correlates to the query;
• comparing the query to a plurality of existing queries and using machine learning to identify similar wording and similar responses;
• using the machine learning to match the query to retrieve and play the answer;
• and
• wherein the interactive device uses a machine learning algorithm to predict a most-likely follow-up query following the query and to retrieve and cache a predicted answer to the follow-up query during a playback of the answer from the recorded interview session, wherein the machine learning algorithm is trained on past patterns of inquiries.

A method of hosting an interactive story-telling conversation between a user and a contributor selected by the user, comprising:

• receiving from the user a selection of the contributor;
• retrieving a first filler recording segment of the contributor, wherein the filler segment features the contributor speaking a pleasantry;
• processing a query from the user, wherein the query is addressed to the contributor;
• comparing the query to a plurality of existing queries and using machine learning for matching the query and for identifying a recording segment featuring the contributor, wherein the recording segment provides an appropriate answer to the query, and wherein the answer is related to the contributor's personal experience or opinion;
• retrieving and playing the identified recording segment to the user;
• using a machine learning algorithm to predict a follow-up query that is most likely to follow the query and to retrieve and cache a predicted recording segment corresponding to a response to the follow-up query during the play of the identified recording segment, wherein the machine learning algorithm is trained on past patterns of inquiries;
• if the user does not pose a follow-up query, playing a second filler segment featuring the contributor in a listening pose;
• and
• if the user poses the predicted follow-up query, processing the follow-up query to identify, retrieve the cached predicted recording segment, and play the cached predicted recording segment that provides an appropriate answer to the follow-up query.