US 12,614,539 B1Granted

Personalized inner voice synthesis using adaptive acoustic parameter modification using demographic data

AssigneeRandolph Ventures LLC

InventorJames Randolph Myers

Life of this patent

Estimating the term…

Loading patent…

Abstract

The systems and methods disclosed herein generate a personalized inner voice audio output that replicates or otherwise shares the acoustic characteristics of a speaker's self-perceived voice, which can differ from their externally perceived voice due to bone conduction effects. The systems and methods disclosed herein can generate a provisional voice clone (e.g., parameters of a voice model) of the speaker's voice (e.g., using a recording of the speaker's voice), and can apply a frequency shift to compensate for the absence of bone-conducted low-frequency adjustment that occurs during natural speech production. A trained artificial intelligence model predicts and applies one or more additional acoustic parameter adjustment values (e.g., formant structure, spectral envelope, prosodic patterns) to the provisional voice clone based on one or more factors (e.g., demographic, anatomical, content, environment). The provisional voice clone can be iteratively refined based on received user feedback (e.g., until the output aligns with the speaker's perception of their inner voice).

Claims20

Dependent claims

A system comprising:

at least one hardware processor;
and
at least one non-transitory memory storing instructions, which, when executed by the at least one hardware processor, cause the system to:
receive (a) audio data comprising one or more speech samples from a speaker and (b) demographic data associated with the speaker,
wherein the audio data represents an external voice of the speaker as captured through air conduction;
input the audio data into a first machine learning model to generate an acoustic parameter set corresponding to a provisional inner voice model,
wherein the provisional inner voice model is configured to synthesize a first speech signal different from the audio data,
wherein the first speech signal of the provisional inner voice model shares one or more acoustic characteristics of the external voice of the speaker;
input the demographic data associated with the speaker into a second machine learning model to generate a first set of transformations,
wherein the second machine learning model is trained on a dataset of acoustic parameters derived from a plurality of users, each associated with (a) corresponding demographic data and (b) one or more corresponding transformations that map a corresponding external voice of the user to a corresponding perceived inner voice of the user, and
wherein the second machine learning model is configured to:
identify a subset of the plurality of users having at least a portion of common demographic data with the speaker, and
generate the first set of transformations using the corresponding transformations associated with the identified subset of the plurality of users;
modify the acoustic parameter set corresponding to the provisional inner voice model by applying the first set of transformations to the acoustic parameter set of the provisional inner voice model to generate a second speech signal,
wherein the second speech signal enables removal of a first set of acoustic deviations between the first speech signal and a perceived inner voice of the speaker;
cause output of the second speech signal generated by the provisional inner voice model at a computing device;
receive, via the computing device, user feedback that defines a second set of transformations configured to remove a second set of acoustic deviations between the second speech signal and the perceived inner voice of the speaker;
and
apply the second set of transformations to the modified acoustic parameter set of the provisional inner voice model.

16 more elements

A non-transitory computer-readable storage medium comprising instructions for generating personalized inner voice audio output stored thereon, wherein the instructions when executed by at least one data processor of a system, cause the system to:

obtain (a) audio data comprising one or more speech samples and (b) demographic data associated with a speaker;
input the audio data into a first artificial intelligence model to generate an acoustic parameter set corresponding to a provisional inner voice model that is configured to synthesize a first speech signal different from the audio data,
wherein the first speech signal of the provisional inner voice model shares one or more acoustic characteristics of an external voice of the speaker;
input the demographic data associated with the speaker into a second artificial intelligence model to generate a first set of transformations,
wherein the second artificial intelligence model is configured to generate the first set of transformations using corresponding transformations associated with one or more users having at least a portion of common demographic data with the speaker;
modify the acoustic parameter set corresponding to the provisional inner voice model by applying the first set of transformations to the acoustic parameter set of the provisional inner voice model to generate a second speech signal,
wherein the second speech signal enables removal of a first set of acoustic deviations between the first speech signal and a perceived inner voice of the speaker;
obtain user feedback that defines a second set of transformations configured to remove a second set of acoustic deviations between the second speech signal and the perceived inner voice of the speaker;
and
apply the second set of transformations to the modified acoustic parameter set of the provisional inner voice model.

7 more elements

A computer-implemented method for generating personalized inner voice audio output, the computer-implemented method comprising:

obtaining audio data comprising one or more speech samples;
inputting the audio data into a first artificial intelligence model to generate an acoustic parameter set corresponding to a provisional inner voice model that is configured to synthesize a first speech signal different from the audio data,
wherein the first speech signal of the provisional inner voice model shares one or more acoustic characteristics of an external voice of a speaker;
determining a first set of transformations for the acoustic parameter set corresponding
to the provisional inner voice model by inputting demographic data associated with the speaker into a second artificial intelligence model to generate the first set of transformations,
wherein the second artificial intelligence model is trained on a dataset of acoustic parameters derived from a plurality of users, each associated with (a) corresponding demographic data and (b) one or more corresponding transformations that map a corresponding external voice of the user to a corresponding perceived inner voice of the user;
causing modification of the acoustic parameter set corresponding to the provisional inner voice model by applying the first set of transformations to the acoustic parameter set of the provisional inner voice model;
obtaining user feedback that defines a second set of transformations configured to remove a set of acoustic deviations between (a) a second speech signal generated by the provisional inner voice model using the modified acoustic parameter set and (b) a particular voice of the speaker;
and
causing application of the second set of transformations to the modified acoustic parameter set of the provisional inner voice model.

7 more elements

Dependencies are inferred from the claim text.

Facts

Application: 19/412,420Filed 8 Dec 2025
Granted: 28 Apr 2026
Term ends: Not estimatedThe source records don’t support an estimate.

Classification10 codes

G10L 13/02
Speech synthesis; Text to speech systems: methods for producing synthetic speech; Speech synthesisers
G10L 13/033
Methods for producing synthetic speech; Speech synthesisers: voice editing, e.g. manipulating the voice of the synthesiser
G10L 21/003
Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its…:…
H04R 3/04
Circuits for transducers: for correcting frequency response