Method and device for processing video, electronic device and storage medium

1. A method for processing a video, comprising:

performing a target recognition to A first video segments and B first speech segments to obtain M second video segments and N second speech segments, wherein the A first video segments and the B first speech segments are obtained by dividing an inputted video, the M second video segments comprise a first face image of a target character, the N second speech segments comprise a first sound of the target character, A is an integer greater than 1, B is a positive integer, M is a positive integer less than or equal to A, and N is a positive integer less than or equal to B;

performing a speech processing to the N second speech segments to obtain N target speech text files, wherein the N second speech segments correspond to the N target speech text files one by one;

extracting first representation information of the first face image from the M second video segments to obtain Q pieces of first representation information, wherein Q is an integer greater than or equal to M;

determining a second sound matched with the target character according to the Q pieces of first representation information; and

merging the second sound with the N target speech text files to obtain N target speech segments, wherein the N target speech text files correspond to the N target speech segments one by one.

2. The method according to claim 1 , wherein performing a speech processing to the N second speech segments to obtain N target speech text files comprises:

performing a speech recognition to the N second speech segments to obtain N text files, wherein the N second speech segments correspond to the N text files one by one; and

translating the N text files according to a specified language to obtain the N target speech text files, wherein the N text files correspond to the N target speech text files one by one.

3. The method according to claim 1 , wherein, extracting first representation information of the first face image from the M second video segments to obtain Q pieces of first representation information comprises:

performing a first representation information extraction to the first face image of each of the M second video segments or the first face image of each of L frames comprising the first face image in the M second video segments, so as to obtain the Q pieces of first representation information, wherein L is a positive integer.

4. The method according to claim 1 , wherein determining a second sound matched with the target character according to the Q pieces of first representation information comprises:

classifying the Q pieces of first representation information to obtain P classes of first representation information, wherein P is a positive integer less than or equal to Q; and

determining the second sound according to one of the P classes with a longest playing period among the inputted video.

5. The method according to claim 1 , wherein, after obtaining the Q pieces of first representation information, the method further comprises:

determining a second face image matched with the target character according to the Q pieces of first representation information;

replacing the first face image in the M second video segments with the second face image to obtain M target video segments, wherein the M video segments correspond to the M target video segments one by one; and

merging the N target speech segments with the M target video segments to obtain an outputted video.

6. The method according to claim 5 , further comprising:

pre-processing the second face image to obtain a third face image;

replacing facial features of the third face image with those of the first image face to obtain a fourth face image;

rectifying the fourth face image with a loss function to obtain a fifth face image;

merging the fifth face image with rest of the target face image except the second face image, to obtain an outputted image.

7. The method according to claim 6 , wherein the pre-processing comprises at least one of a face alignment, an image enhancement and a normalization.

8. The method according to claim 1 , wherein, before performing a target recognition to A first video segments and B first speech segments, the method further comprises:

dividing the inputted video into the A first video segments according to a preset period or a playing period of the inputted video; and

dividing the inputted video into the B first speech segments according to a preset volume threshold.

9. The method according to claim 1 , further comprising:

extracting second representation information of the first sound from the N second speech segments to obtain R pieces of second representation information, wherein R is an integer greater than or equal to N; and

determining the second sound according to the Q pieces of first representation information and the R pieces of second representation information.

10. An electronic device, comprising: a housing, a processor, a memory, a circuit board and a power supply circuit; wherein the circuit board is enclosed by the housing; the processor and the memory are positioned on the circuit board; the power supply circuit is configured to provide power for respective circuits or components of the electronic device; the memory is configured to store executable program codes; and the processor is configured to run a program corresponding to the executable program codes by reading the executable program codes stored in the memory, to perform a method for processing a video, the method comprising:

performing a target recognition to A first video segments and B first speech segments to obtain M second video segments and N second speech segments, wherein the A first video segments and the B first speech segments are obtained by dividing an inputted video, the M second video segments comprise a first face image of a target character, the N second speech segments comprise a first sound of the target character, A is an integer greater than 1, B is a positive integer, M is a positive integer less than or equal to A, and N is a positive integer less than or equal to B;

performing a speech processing to the N second speech segments to obtain N target speech text files, wherein the N second speech segments correspond to the N target speech text files one by one;

extracting first representation information of the first face image from the M second video segments to obtain Q pieces of first representation information, wherein Q is an integer greater than or equal to M;

determining a second sound matched with the target character according to the Q pieces of first representation information; and

merging the second sound with the N target speech text files to obtain N target speech segments, wherein the N target speech text files correspond to the N target speech segments one by one.

11. The electronic device according to claim 10 , wherein the processor is configured to perform a speech processing to the N second speech segments to obtain N target speech text files by acts of:

performing a speech recognition to the N second speech segments to obtain N text files, wherein the N second speech segments correspond to the N text files one by one; and

translating the N text files according to a specified language to obtain the N target speech text files, wherein the N text files correspond to the N target speech text files one by one.

12. The electronic device according to claim 10 , wherein the processor is configured to extract first representation information of the first face image from the M second video segments to obtain Q pieces of first representation information by acts of:

performing a first representation information extraction to the first face image of each of the M second video segments or the first face image of each of L frames comprising the first face image in the M second video segments, so as to obtain the Q pieces of first representation information, wherein L is a positive integer.

13. The electronic device according to claim 10 , wherein the processor is configured to determine a second sound matched with the target character according to the Q pieces of first representation information by acts of:

classifying the Q pieces of first representation information to obtain P classes of first representation information, wherein P is a positive integer less than or equal to Q; and

determining the second sound according to one of the P classes with a longest playing period among the inputted video.

14. The electronic device according to claim 10 , wherein the processor is configured to, after obtaining the Q pieces of first representation information, perform acts of:

determining a second face image matched with the target character according to the Q pieces of first representation information;

replacing the first face image in the M second video segments with the second face image to obtain M target video segments, wherein the M video segments correspond to the M target video segments one by one; and

merging the N target speech segments with the M target video segments to obtain an outputted video.

15. The electronic device according to claim 14 , wherein processor is configured to perform acts of:

pre-processing the second face image to obtain a third face image;

replacing facial features of the third face image with those of the first image face to obtain a fourth face image;

rectifying the fourth face image with a loss function to obtain a fifth face image;

merging the fifth face image with rest of the target face image except the second face image, to obtain an outputted image.

16. The electronic device according to claim 15 , wherein the pre-processing comprises at least one of a face alignment, an image enhancement and a normalization.

17. The electronic device according to claim 10 , wherein the processor is configured to, before performing a target recognition to A first video segments and B first speech segments, perform acts of:

dividing the inputted video into the A first video segments according to a preset period or a playing period of the inputted video; and

dividing the inputted video into the B first speech segments according to a preset volume threshold.

18. The electronic device according to claim 10 , wherein the processor is configured to perform acts of:

extracting second representation information of the first sound from the N second speech segments to obtain R pieces of second representation information, wherein R is an integer greater than or equal to N; and

determining the second sound according to the Q pieces of first representation information and the R pieces of second representation information.

19. A non-transitory computer readable storage medium, having computer programs stored therein, when the computer programs are executed by a processor, a method for processing a video is realized, the method comprising:

performing a target recognition to A first video segments and B first speech segments to obtain M second video segments and N second speech segments, wherein the A first video segments and the B first speech segments are obtained by dividing an inputted video, the M second video segments comprise a first face image of a target character, the N second speech segments comprise a first sound of the target character A is an integer greater than 1, B is a positive integer, M is a positive integer less than or equal to A, and N is a positive integer less than or equal to B;

performing a speech processing to the N second speech segments to obtain N target speech text files, wherein the N second speech segments correspond to the N target speech text files one by one;

extracting first representation information of the first face image from the M second video segments to obtain Q pieces of first representation information, wherein Q is an integer greater than or equal to M;

determining a second sound matched with the target character according to the Q pieces of first representation information; and

merging the second sound with the N target speech text files to obtain N target speech segments, wherein the N target speech text files correspond to the N target speech segments one by one.

20. The non-transitory computer readable storage medium according to claim 19 , wherein performing a speech processing to the N second speech segments to obtain N target speech text files comprises:

performing a speech recognition to the N second speech segments to obtain N text files, wherein the N second speech segments correspond to the N text files one by one; and

translating the N text files according to a specified language to obtain the N target speech text files, wherein the N text files correspond to the N target speech text files one by one.