Methods, systems and apparatuses for inner voice recovery from neural activation relating to sub-vocalization

A method for sub-vocalization, the method comprising:

• performing, by the processor that executes instructions contained in a non-transitory computer read-able medium, a continuous-to-continuous mapping of neural signal data received from one or more body sensors connected to an user wherein the one or more body sensors monitors at least neural activities of the user of a sub-vocalized voice at a sensory level and sends the neural signal data to the processor;
• receiving, by the processor, the neural signal data in an iterative closed loop for training the processor wherein the training enables generating a sufficiently large data set in the neural signal domain from the neural signal data to link to a produced voice domain;
• constructing a common feature space, by the processor, which associates the neural signal domain with the produced voice domain for producing audio communications wherein the common feature space implicitly extracts features related to audio communications using the sufficiently large data set by the common feature space for linking neural signal domain data to the produced voice data without requiring any prior feature classification of the received neural signal data;
• modeling by artificial neural networks (ANN) the continuous-to-continuous mapping of neural signal data received from the one or more body sensors connected to the user;
• learning, by the processor, by iterations of the closed loop training of the processor a mapping of the common feature space for a set of common neural signal domain activity and, by applying a set of ANN based algorithms of the ANN while simultaneously modeling the common feature space;
• mapping by a first function the neural signal domain, and targeting by a second function the mapping of the first function to the produced voice domain to reproduce the subvocalized voice;
• and
• using a f(x) by the first function for mapping to the common feature space, and using an inverse G−1(x) by the second function for targeting to the produced voice domain.

A sub-vocalization system of imagined speech, comprising:

• a plurality of body sensors;
• a processor that executes instructions contained in a non-transitory computer read-able medium, to perform a continuous-to-continuous mapping of neural signal data received from one or more body sensors from the plurality of body sensors connected to an user wherein the one or more body sensors monitors at least neural activities of the user of a sub-vocalized voice at a sensory level and sends the neural signal data to the processor;
• and
• an artificial neural network (ANN) to model the continuous-to-continuous mapping of neural signal data received from the one or more body sensors connected to the user, the processor further configured to:
• process, a set of neural signal data received from the one or more body sensors, in an iterative closed loop for training the processor wherein the training generates a sufficiently large data set in the neural signal domain from the neural signal data to link to a produced voice domain;
• construct, a common feature space, which associates the neural signal domain with the produced voice domain to produce audio communications wherein the common feature space implicitly extracts features related to audio communications, using the sufficiently large data set, by the common feature space to link neural signal domain data to the produced voice data without requiring any prior feature classification of the received neural signal data;
• and
• learn, using a data learning, by performing iterations of the closed loop training of the processor of a mapping of the common feature space for a set of common neural signal domain activity, and by applying a set of ANN based algorithms of the ANN while simultaneously performing a data modeling of the common feature space.

A sub-vocalization system of imagined speech, comprising:

• a plurality of body sensors;
• a processor, that executes instructions contained in a non-transitory computer read-able medium, to perform a continuous-to-continuous mapping of neural signal data received from one or more body sensors from the plurality of body sensors connected to an user wherein the one or more body sensors monitors at least neural activities of the user of a sub-vocalized voice at a sensory level and sends the neural signal data to the processor;
• and
• an artificial neural network (ANN) to model the continuous-to-continuous mapping of neural signal data received from the one or more body sensors connected to the user, the processor configured to:
• process, a set of neural signal data received from the one or more body sensors, in an iterative closed loop for training the processor wherein the training generates a sufficiently large data set in the neural signal domain from the neural signal data to link to a produced voice domain;
• construct, a common feature space, which associates the neural signal domain with the produced voice domain to produce audio communications wherein the common feature space implicitly extracts features related to audio communications, using the sufficiently large data set, by the common feature space to link neural signal domain data to the produced voice data without requiring any prior feature classification of the received neural signal data;
• learn, using a data learning, by performing iterations of the closed loop training of the processor of a mapping of the common feature space for a set of common neural signal domain activity, and by applying a set of ANN based algorithms of the ANN while performing a data modeling of the common feature space:
• and
• perform a first function to map the neural signal domain, and a second function to target the map of the first function to the produced voice domain to reproduce the subvocalized voice wherein a f(x) is used by the first function to map to the common feature space, and an inverse G-1(x) is used by the second function for targeting to the produced voice domain.