Accent personalization for speakers and listeners
In one aspect, an example methodology implementing the disclosed techniques includes, by a computing device, receiving audio data corresponding to a spoken utterance by a first user and determining an accent of the audio data. The method also includes, by the computing device, neutralizing the accent of the audio data to a preconfigured accent and transmitting a modified audio data in the preconfigured accent to another computing device. The modified audio data includes the spoken utterance by the first user.
1 . A method comprising:
providing a service on a computing device and an application on another computing device;
receiving, by a computing device, audio data corresponding to a spoken utterance by a first user;
determining, by the computing device, an accent of the audio data using an accent classification module, the accent classification module comprising a convolution layer, a maximum pooling layer, and a softmax layer;
selecting a neutralization model from a plurality of models, each of the plurality of models trained to neutralize a different accent;
neutralizing, by the computing device, the accent of the audio data to generate a modified audio data in a preconfigured accent using the selected neutralization model;
transmitting, by the computing device, the modified audio data in the preconfigured accent to the another computing device being used by a second user;
personalizing, by the application, the modified audio data to a first accent selected by the second user;
personalizing, by the application, the voice of the spoken utterance to a second voice selected by the second user; and
outputting, by the application, the cloned voice to an audio playback device.
2 . The method of claim 1 , wherein the preconfigured accent is a native English accent.
3 . The method of claim 1 , wherein the accent of the audio data is a non-native English accent.
4 . The method of claim 1 , wherein the accent of the audio data is a native English accent.
5 . The method of claim 1 , wherein the accent of the second user is a native English accent.
6 . The method of claim 1 , wherein the accent of the second user is a non-native English accent.
7 . The method of claim 1 , further comprising:
wherein the second voice is selected to be one of a voice of the first user and a voice of the second user.
8 . The method of claim 7 , wherein the second voice is a cloning of the voice of the second user.
9 . A computing device comprising:
one or more non-transitory machine-readable mediums configured to store instructions; and
one or more processors configured to execute the instructions stored on the one or more non-transitory machine-readable mediums, wherein execution of the instructions causes the one or more processors to carry out a process comprising:
providing a service on a computing device and an application on another computing device;
receiving audio data corresponding to a spoken utterance by a first user;
determining, by the computing device, an accent of the audio data using an accent classification module, the accent classification module comprising a convolution layer, a maximum pooling layer, and a softmax layer;
selecting a neutralization model from a plurality of models, each of the plurality of models trained to neutralize a different accent;
neutralizing, by the computing device, the accent of the audio data to generate a modified audio data in a preconfigured accent using the selected neutralization model;
transmitting, by the computing device, the modified audio data in the preconfigured accent to the another computing device being used by a second user
personalizing, by the application, the modified audio data to a first accent selected by the second user;
personalizing, by the application, the voice of the spoken utterance to a second voice selected by the second user; and
outputting, by the application, the cloned voice to an audio playback device.
10 . The computing device of claim 9 , wherein the preconfigured accent is a native English accent.
11 . The computing device of claim 9 , wherein the accent of the audio data is a non-native English accent.
12 . The computing device of claim 9 , wherein the accent of the audio data is a native English accent.
13 . The computing device of claim 9 , wherein the accent of the second user is a native English accent.
14 . The computing device of claim 9 , wherein the accent of the second user is a non-native English accent.
15 . The computing device of claim 9 , wherein the second voice is selected to be one of a voice of the first user and a voice of the second user.
16 . The computing device of claim 15 , wherein the second voice is a cloning of the voice of the second user.
17 . The computing device of claim 15 , wherein the another voice is a voice specified by the second user.
18 . A non-transitory machine-readable medium encoding instructions that when executed by one or more processors cause a process to be carried out, the process including:
providing a service on a computing device and an application on another computing device;
receiving audio data corresponding to a spoken utterance by a first user;
determining an accent of the audio data using an accent classification module, the accent classification module comprising a convolution layer, a maximum pooling layer, and a softmax layer;
selecting a neutralization model from a plurality of models, each of the plurality of models trained to neutralize a different accent;
neutralizing the accent of the audio data to generate a modified audio data in a preconfigured accent using the selected neutralization model;
transmitting the modified audio data in the preconfigured accent to the another computing device being used by a second user;
personalizing the modified audio data to a first accent selected by the second user;
personalizing the voice of the spoken utterance to a second voice selected by the second user; and
outputting the cloned voice to an audio playback device.