System and method for synthesizing spoken dialogue using deep learning

Abstract

Disclosed are a system, method and software to associate attributes with digital media assets using deep learning. Digital media contains specific assets, such as audio, that can be replaced with other assets. The system, method and software allow for personalizing digital media content based in part on neural network analysis in order to generate synthesized audio, such as speech, spoken prompts, and/or dialogue.

What is claimed is: 
     
         1 . A method for personalizing media content using synthesized audio, comprising:
 generating, by a server, a script configured to output a digital media stream;   capturing, by a computing device communicatively coupled to the server, a user interaction with the digital media stream;   analyzing, by the server, the user interaction to determine a user affinity for the digital media stream;   selecting, by the server, an audio asset based on the user affinity; and   updating, by the server, the script to insert the audio asset into the digital media stream.   
     
     
         2 . The method of  claim 1 , wherein the audio asset is a speech-based audio asset. 
     
     
         3 . The method of  claim 1 , wherein the audio asset is a song. 
     
     
         4 . The method of  claim 1 , wherein the user affinity is selected from a group comprising at least one of a preference, an emotional value, a cognitive value, and a social value. 
     
     
         5 . The method of  claim 1 , wherein the user interaction is selected from a group comprising at least one of a viewing habit, a purchase, a selection, and an answer to a question. 
     
     
         6 . The method of  claim 1 , wherein the selecting utilizes an artificial intelligence method to identify the digital audio asset that matches the user affinity. 
     
     
         7 . The method of  claim 1 , wherein the selecting utilizes a neural network to identify the digital audio asset that matches the user affinity. 
     
     
         8 . The method of  claim 1 , wherein the user interaction is an input indicating a like or dislike by a user. 
     
     
         9 . A system for personalizing media content using synthesized audio, comprising:
 a content distributor configured to generate a script, the script configured to output a digital media stream;   a server coupled to the content distributor, the server configured to capture a user interaction with the digital media stream; and   a correlation algorithm communicatively coupled to the server, the correlation algorithm configured to correlate a user affinity with the user interaction,   wherein the content distributor is further configured to update the script to insert an audio asset into the digital media stream, wherein the audio asset is selected based on the user affinity.   
     
     
         10 . The system of  claim 9 , wherein the audio asset is a speech-based audio asset. 
     
     
         11 . The system of  claim 9 , wherein the audio asset is a spoken prompt. 
     
     
         12 . The system of  claim 9 , wherein the user affinity is selected from a group comprising at least one of a preference, an emotional value, a cognitive value, and a social value. 
     
     
         13 . The system of  claim 9 , wherein the user interaction is selected from a group comprising at least one of a viewing habit, a purchase, a selection, and an answer to a question. 
     
     
         14 . The system of  claim 9 , wherein the content distributor utilizes an artificial intelligence method to select the digital audio asset that matches the user affinity. 
     
     
         15 . The system of  claim 9 , wherein the content distributor utilizes a neural network to select the digital audio asset that matches the user affinity. 
     
     
         16 . A method for providing interactive synthesized audio content, comprising:
 generating, by a server, a script to output a digital audio stream in the form of a spoken question;   capturing, by a computing device communicatively coupled to the server, a user response to the spoken question;   selecting, by the server, an audio asset based on the user response;   updating, by the server, the script to insert the audio asset, thereby generating a second digital audio stream; and   outputting to the computing device, by the server, the second digital audio stream.   
     
     
         17 . The method of  claim 16 , wherein the user response is in the form of audio captured from a user. 
     
     
         18 . The method of  claim 16 , wherein the user response is the form of an emotion captured from a user. 
     
     
         19 . The method of  claim 16 , wherein the user response is an input indicating a like or dislike by a user. 
     
     
         20 . The method of  claim 16 , wherein the selecting utilizes an artificial intelligence method to identify the audio asset based on the user response.