Audio processing method and apparatus, device, readable storage medium, and program product

1 . An audio processing method performed by a computer device, the method comprising:

obtaining a piece of voice data captured by a microphone of a terminal associated with a game account;

segmenting the piece of voice data into a plurality of audio segments;

determining a voiceprint vector corresponding to each of the plurality of audio segments, wherein each of the plurality of audio segments is generated by a respective audio object;

determining an initial similarity matrix according to the voiceprint vectors corresponding to the plurality of audio segments, each element in the initial similarity matrix comprising a similarity between voiceprint vectors corresponding to a respective pair of audio segments;

for a respective row of the initial similarity matrix,

sorting similarities in the row according to an increasing order to obtain a first sorting result;

determining a similarity difference between two adjacent similarities in the first sorting result to obtain a plurality of similarity differences;

determining a maximum similarity difference in the plurality of similarity differences, the maximum similarity difference having a pair a minuend and a subtrahend;

determining the minuend of the maximum similarity difference as a dynamic threshold corresponding to the row;

adjusting the initial similarity matrix according to the dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix, further comprising:

setting a similarity in a k th row of the initial similarity matrix that is less than a dynamic threshold corresponding to the k th row to zero, k being a positive integer; and

obtaining the reference similarity matrix based on an adjusted result of each row, k being a positive integer;

determining, according to the reference similarity matrix, a quantity of audio objects corresponding to the plurality of audio segments, wherein the quantity of audio objects is no greater than a quantity of the plurality of audio segments, further including:

for a respective reference parameter p of a plurality of reference parameters, p being a positive integer,

calculating a first quantity of first eigenvalues for the reference parameter p;

determining a difference between two adjacent first eigenvalues of the first quantity of first eigenvalues to obtain a plurality of eigenvalue differences;

calculating a ratio value between the reference parameter p and a maximum of the plurality of eigenvalue differences;

determining the quantity of the audio objects in the plurality of audio segments as one of the plurality of reference parameters when its corresponding ratio value is minimized;

clustering the plurality of audio segments according to the quantity of the audio objects to obtain an audio segment corresponding to each audio object; and

determining a unique user associated with the game account based on the audio segment corresponding to each audio object.

2 . The method according to claim 1 , wherein the adjusting the initial similarity matrix according to a dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix comprises:

multiplying a similarity that is in similarities comprised in a k th row of the initial similarity matrix and that is less than a dynamic threshold corresponding to the k th row by a second value, and obtaining the reference similarity matrix based on an adjusted result of each row.

3 . The method according to claim 1 , wherein each voiceprint vector represents a voiceprint feature corresponding to a respective audio segment.

4 . The method according to claim 1 , wherein the dynamic threshold is used for adjusting similarity differences between different voiceprint vectors corresponding to a respective pair of audio segments within the row of the initial similarity matrix.

5 . A computer device, comprising a processor and a memory, the memory having at least one program code stored therein, and the at least one program code being loaded and executed by the processor to enable the computer device to implement an audio processing method including:

obtaining a piece of voice data captured by a microphone of a terminal associated with a game account;

segmenting the piece of voice data into a plurality of audio segments;

determining a voiceprint vector corresponding to each of the plurality of audio segments, wherein each of the plurality of audio segments is generated by a respective audio object;

determining an initial similarity matrix according to the voiceprint vectors corresponding to the plurality of audio segments, each element in the initial similarity matrix comprising a similarity between voiceprint vectors corresponding to a respective pair of audio segments;

for a respective row of the initial similarity matrix,

sorting similarities in the row according to an increasing order to obtain a first sorting result;

determining a similarity difference between two adjacent similarities in the first sorting result to obtain a plurality of similarity differences;

determining a maximum similarity difference in the plurality of similarity differences, the maximum similarity difference having a pair a minuend and a subtrahend;

determining the minuend of the maximum similarity difference as a dynamic threshold corresponding to the row;

adjusting the initial similarity matrix according to the dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix, further comprising:

setting a similarity in a k th row of the initial similarity matrix that is less than a dynamic threshold corresponding to the k th row to zero, k being a positive integer; and

obtaining the reference similarity matrix based on an adjusted result of each row, k being a positive integer;

determining, according to the reference similarity matrix, a quantity of audio objects corresponding to the plurality of audio segments, wherein the quantity of audio objects is no greater than a quantity of the plurality of audio segments, further including:

for a respective reference parameter p of a plurality of reference parameters, p being a positive integer,

calculating a first quantity of first eigenvalues for the reference parameter p;

determining a difference between two adjacent first eigenvalues of the first quantity of first eigenvalues to obtain a plurality of eigenvalue differences;

calculating a ratio value between the reference parameter p and a maximum of the plurality of eigenvalue differences;

determining the quantity of the audio objects in the plurality of audio segments as one of the plurality of reference parameters when its corresponding ratio value is minimized;

clustering the plurality of audio segments according to the quantity of the audio objects to obtain an audio segment corresponding to each audio object; and

determining a unique user associated with the game account based on the audio segment corresponding to each audio object.

6 . The computer device according to claim 5 , wherein the adjusting the initial similarity matrix according to a dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix comprises:

multiplying a similarity that is in similarities comprised in a k th row of the initial similarity matrix and that is less than a dynamic threshold corresponding to the k th row by a second value, and obtaining the reference similarity matrix based on an adjusted result of each row.

7 . The computer device according to claim 5 , wherein each voiceprint vector represents a voiceprint feature corresponding to a respective audio segment.

8 . The computer device according to claim 5 , wherein the dynamic threshold is used for adjusting similarity differences between different voiceprint vectors corresponding to a respective pair of audio segments within the row of the initial similarity matrix.

9 . A non-transitory computer-readable storage medium, having at least one program code stored thereon, the at least one program code being loaded and executed by a processor of a computer device to cause the computer device to implement an audio processing method including:

obtaining a piece of voice data captured by a microphone of a terminal associated with a game account;

segmenting the piece of voice data into a plurality of audio segments;

determining a voiceprint vector corresponding to each of the plurality of audio segments, wherein each of the plurality of audio segments is generated by a respective audio object;

determining an initial similarity matrix according to the voiceprint vectors corresponding to the plurality of audio segments, each element in the initial similarity matrix comprising a similarity between voiceprint vectors corresponding to a respective pair of audio segments;

for a respective row of the initial similarity matrix,

sorting similarities in the row according to an increasing order to a first order to obtain a first sorting result;

determining a similarity difference between two adjacent similarities in the first sorting result to obtain a plurality of similarity differences;

determining a maximum similarity difference in the plurality of similarity differences, the maximum similarity difference having a pair a minuend and a subtrahend;

determining the minuend of the maximum similarity difference as a dynamic threshold corresponding to the row;

adjusting the initial similarity matrix according to the dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix, further comprising:

setting a similarity in a k th row of the initial similarity matrix that is less than a dynamic threshold corresponding to the k th row to zero, k being a positive integer; and

obtaining the reference similarity matrix based on an adjusted result of each row, k being a positive integer;

determining, according to the reference similarity matrix, a quantity of audio objects corresponding to the plurality of audio segments, wherein the quantity of audio objects is no greater than a quantity of the plurality of audio segments, further including:

for a respective reference parameter p of a plurality of reference parameters, p being a positive integer,

calculating a first quantity of first eigenvalues for the reference parameter p;

determining a difference between two adjacent first eigenvalues of the first quantity of first eigenvalues to obtain a plurality of eigenvalue differences;

calculating a ratio value between the reference parameter p and a maximum of the plurality of eigenvalue differences;

determining the quantity of the audio objects in the plurality of audio segments as one of the plurality of reference parameters when its corresponding ratio value is minimized;

clustering the plurality of audio segments according to the quantity of the audio objects to obtain an audio segment corresponding to each audio object; and

determining a unique user associated with the game account based on the audio segment corresponding to each audio object.

10 . The non-transitory computer-readable storage medium according to claim 9 , wherein the adjusting the initial similarity matrix according to a dynamic threshold corresponding to each row of the initial similarity matrix to obtain a reference similarity matrix comprises:

multiplying a similarity that is in similarities comprised in a k th row of the initial similarity matrix and that is less than a dynamic threshold corresponding to the k th row by a second value, and obtaining the reference similarity matrix based on an adjusted result of each row.

11 . The non-transitory computer-readable storage medium according to claim 9 , wherein the dynamic threshold is used for adjusting similarity differences between different voiceprint vectors corresponding to a respective pair of audio segments within the row of the initial similarity matrix.