CN117440001A - Data synchronization method based on message - Google Patents

Data synchronization method based on message Download PDF

Info

Publication number
CN117440001A
CN117440001A CN202311753182.5A CN202311753182A CN117440001A CN 117440001 A CN117440001 A CN 117440001A CN 202311753182 A CN202311753182 A CN 202311753182A CN 117440001 A CN117440001 A CN 117440001A
Authority
CN
China
Prior art keywords
voice
data
original
frame
voice message
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202311753182.5A
Other languages
Chinese (zh)
Other versions
CN117440001B (en
Inventor
李盼盼
王子佩
甘海鹏
邓诗雨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sdic Human Resources Service Co ltd
Original Assignee
Sdic Human Resources Service Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sdic Human Resources Service Co ltd filed Critical Sdic Human Resources Service Co ltd
Priority to CN202311753182.5A priority Critical patent/CN117440001B/en
Publication of CN117440001A publication Critical patent/CN117440001A/en
Application granted granted Critical
Publication of CN117440001B publication Critical patent/CN117440001B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1095—Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Computational Linguistics (AREA)
  • Quality & Reliability (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Telephonic Communication Services (AREA)

Abstract

The invention discloses a data synchronization method based on information, which belongs to the technical field of data processing and comprises the following steps: s1, acquiring original voice data, preprocessing the original voice data to generate de-noised voice data, inputting the original voice data and the de-noised voice data into a constructed voice processing model, and generating a voice message queue; s2, determining voice message weight according to the voice message queue, and generating a voice message report; s3, synchronously transmitting the voice message report and the denoising voice data to the user terminal. The invention synchronizes the preprocessed voice data and voice message report to the user terminal, such as a mobile phone, so that the user can conveniently and synchronously check the voice data in time, and acquire the audio change condition of the voice message according to the voice message report, thereby quickly knowing the audio condition.

Description

Data synchronization method based on message
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to a data synchronization method based on a message.
Background
With the improvement of economy and technology, the development and popularization of mobile internet technology, data transmission technology and intelligent terminals, and the development of audio transmission are widely progressed. The requirements of people on the audio quality are gradually improved, so that the research on the quality improvement of videos has very important significance for enriching the mental and living demands of people. At present, with the development of audio collection technology, how to synchronize specific information of audio collected by an audio collection device (such as a microphone) and a voice signal thereof to a user terminal (such as a mobile phone) becomes an urgent problem to be solved.
Disclosure of Invention
The invention provides a data synchronization method based on a message in order to solve the problems.
The technical scheme of the invention is as follows: a message-based data synchronization method comprising the steps of:
s1, acquiring original voice data, preprocessing the original voice data to generate de-noised voice data, inputting the original voice data and the de-noised voice data into a constructed voice processing model, and generating a voice message queue;
s2, determining voice message weight according to the voice message queue, and generating a voice message report;
s3, synchronously transmitting the voice message report and the denoising voice data to the user terminal.
Further, S1 comprises the following sub-steps:
s11, collecting a plurality of frames of original voice data, denoising each frame of original voice data, and generating each frame of denoised voice data;
s12, constructing a voice processing model, inputting original voice data of each frame and denoising voice data of each frame into the voice processing model, and generating a voice message sequence;
s13, arranging all voice message values in the voice message sequence according to time sequence as a voice message queue.
The beneficial effects of the above-mentioned further scheme are: in the invention, the voice processing model can be used for extracting the audio characteristic values of the original voice data and the denoising voice data, carrying out difference operation and absolute value operation on the audio characteristic values of each frame, and fusing the operated audio characteristic values, so that the audio characteristics can be enriched, and the voice message values output by the characteristic generation layer can contain more voice characteristics after time sequencing.
Further, in S12, the speech processing model includes a first feature extraction layer, a second feature extraction layer, an operator A1, a feature fusion layer, and a feature generation layer;
the input end of the first feature extraction layer and the input end of the second feature extraction layer are both used as input ends of a voice processing model; the output end of the first characteristic extraction layer is connected with the first input end of the arithmetic unit A1; the output end of the second characteristic extraction layer is connected with the second input end of the arithmetic unit A1; the output end of the arithmetic unit A1 is connected with the input end of the feature fusion layer; the output end of the characteristic fusion layer is connected with the input end of the characteristic generation layer; the output of the feature generation layer serves as the output of the speech processing model.
Further, the expression of the first feature extraction layer is: x= { X 1 ,x 2 ,..,x n ,..,x N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, X represents the output of the first feature extraction layer, N represents the total number of frames of the original speech data, X n An audio feature value representing the original speech data of the nth frame, s representing the pitch of the original speech data, f representing the sampling frequency of the original speech data, l n Indicating the loudness of the original voice data of the nth frame, and e indicating an index;
the expression of the second feature extraction layer is: y= { Y 1 ,y 2 ,..,y m ,..,y M ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where m=1, 2, M, Y denote the output of the second feature extraction layer, M denote the total number of frames of the denoised speech data, Y m Audio characteristic value representing m-th frame of de-noised speech data,/->Pitch representing denoised speech data, l m Representing the loudness of the m-th frame of de-noised speech data.
Further, operation AThe expression of 1 is: z= { x 1 -y 1 ,x 2 -y 2 ,..,x n -y m ,..,x N -y M N=m; where n=1, 2, N, m=1, 2, M, x n Audio feature value, y, representing n-th frame of original speech data m The audio characteristic value of the M-th frame of de-noised voice data is represented, N represents the total frame number of the original voice data, and M represents the total frame number of the de-noised voice data.
Further, the expression of the feature fusion layer is:the method comprises the steps of carrying out a first treatment on the surface of the Wherein P represents the output of the feature fusion layer, σ k A weight parameter, b, representing the kth neuron in the feature fusion layer k The bias parameter of the kth neuron in the feature fusion layer is represented, Z represents the output of an operator A1, K represents the number of neurons in the feature fusion layer, max (DEG) represents the maximum value operation, X represents the output of the first feature extraction layer, and Y represents the output of the second feature extraction layer.
Further, the expression of the feature generation layer is: r= { R 1 ,r 2 ,..,r n ,..,r N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, N represents the total number of frames of the original speech data, r n Voice message value, x, representing n-th frame of original voice data n An audio feature value representing the original speech data of the nth frame, and P representing the output of the feature generation layer.
Further, S2 comprises the following sub-steps:
s21, acquiring a voice message blank report;
s22, calculating the voice message weight of each frame of original voice data according to the voice message queue to generate a voice message weight set;
s23, performing de-duplication processing on the voice message weight set to generate standard voice message weight of each frame of original voice data;
s23, generating a voice message report according to the standard voice message weight of the original voice data of each frame.
The beneficial effects of the above-mentioned further scheme are: in the invention, the weight operation is carried out on the voice message queue containing a plurality of voice signal characteristics, the voice message weight is filled into the blank report form according to the sequence from large to small, the position with larger weight indicates that the change of the audio signal is larger or more important, and the user can conveniently and rapidly obtain the important information of the voice data when viewing the voice message report form.
Further, in S22, the voice message weight ρ of the n-th frame of original voice data n The calculation formula of (2) is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein r is n Voice message value r representing n-th frame of original voice data n+1 A voice message value representing the n+1th frame of original voice data, epsilon represents a minimum value, and N represents the total frame number of the original voice data.
Further, in S24, the specific method for generating the voice message report includes: and filling the standard voice message weight of each frame of original voice data into the voice message blank report according to the sequence from large to small to generate a voice message report.
The beneficial effects of the invention are as follows: the invention discloses a data synchronization method based on messages, which is characterized in that voice data is preprocessed to generate a voice message queue containing a large number of audio features, and after the weight operation is carried out on the voice message queue, a voice message blank report is filled in to generate a final voice message report; finally, the invention synchronizes the preprocessed voice data and voice message report to the user terminal, such as mobile phone, so that the user can check the voice data in time and acquire the audio change condition of the voice message according to the voice message report, and the audio condition can be known quickly.
Drawings
FIG. 1 is a flow chart of a message-based data synchronization method;
fig. 2 is a schematic diagram of a speech processing model.
Detailed Description
Embodiments of the present invention are further described below with reference to the accompanying drawings.
As shown in fig. 1, the present invention provides a message-based data synchronization method, which includes the following steps:
s1, acquiring original voice data, preprocessing the original voice data to generate de-noised voice data, inputting the original voice data and the de-noised voice data into a constructed voice processing model, and generating a voice message queue;
s2, determining voice message weight according to the voice message queue, and generating a voice message report;
s3, synchronously transmitting the voice message report and the denoising voice data to the user terminal.
In an embodiment of the present invention, S1 comprises the following sub-steps:
s11, collecting a plurality of frames of original voice data, denoising each frame of original voice data, and generating each frame of denoised voice data;
s12, constructing a voice processing model, inputting original voice data of each frame and denoising voice data of each frame into the voice processing model, and generating a voice message sequence;
s13, arranging all voice message values in the voice message sequence according to time sequence as a voice message queue.
In the invention, the voice processing model can be used for extracting the audio characteristic values of the original voice data and the denoising voice data, carrying out difference operation and absolute value operation on the audio characteristic values of each frame, and fusing the operated audio characteristic values, so that the audio characteristics can be enriched, and the voice message values output by the characteristic generation layer can contain more voice characteristics after time sequencing.
In the embodiment of the present invention, as shown in fig. 2, in S12, the speech processing model includes a first feature extraction layer, a second feature extraction layer, an operator A1, a feature fusion layer, and a feature generation layer;
the input end of the first feature extraction layer and the input end of the second feature extraction layer are both used as input ends of a voice processing model; the output end of the first characteristic extraction layer is connected with the first input end of the arithmetic unit A1; the output end of the second characteristic extraction layer is connected with the second input end of the arithmetic unit A1; the output end of the arithmetic unit A1 is connected with the input end of the feature fusion layer; the output end of the characteristic fusion layer is connected with the input end of the characteristic generation layer; the output of the feature generation layer serves as the output of the speech processing model.
In the embodiment of the present invention, the expression of the first feature extraction layer is: x= { X 1 ,x 2 ,..,x n ,..,x N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, X represents the output of the first feature extraction layer, N represents the total number of frames of the original speech data, X n An audio feature value representing the original speech data of the nth frame, s representing the pitch of the original speech data, f representing the sampling frequency of the original speech data, l n Indicating the loudness of the original voice data of the nth frame, and e indicating an index;
the expression of the second feature extraction layer is: y= { Y 1 ,y 2 ,..,y m ,..,y M ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where m=1, 2, M, Y denote the output of the second feature extraction layer, M denote the total number of frames of the denoised speech data, Y m Audio characteristic value representing m-th frame of de-noised speech data,/->Pitch representing denoised speech data, l m Representing the loudness of the m-th frame of de-noised speech data.
In the embodiment of the present invention, the expression of the operation A1 is: z= { x 1 -y 1 ,x 2 -y 2 ,..,x n -y m ,..,x N -y M N=m; where n=1, 2, N, m=1, 2, M, x n Audio feature value, y, representing n-th frame of original speech data m The audio characteristic value of the M-th frame of de-noised voice data is represented, N represents the total frame number of the original voice data, and M represents the total frame number of the de-noised voice data.
In the embodiment of the invention, the expression of the feature fusion layer is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein P represents the output of the feature fusion layer, σ k A weight parameter, b, representing the kth neuron in the feature fusion layer k The bias parameter of the kth neuron in the feature fusion layer is represented, Z represents the output of an operator A1, K represents the number of neurons in the feature fusion layer, max (DEG) represents the maximum value operation, X represents the output of the first feature extraction layer, and Y represents the output of the second feature extraction layer.
In the embodiment of the present invention, the expression of the feature generation layer is: r= { R 1 ,r 2 ,..,r n ,..,r N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, N represents the total number of frames of the original speech data, r n Voice message value, x, representing n-th frame of original voice data n An audio feature value representing the original speech data of the nth frame, and P representing the output of the feature generation layer.
In an embodiment of the present invention, S2 comprises the following sub-steps:
s21, acquiring a voice message blank report;
s22, calculating the voice message weight of each frame of original voice data according to the voice message queue to generate a voice message weight set;
s23, performing de-duplication processing on the voice message weight set to generate standard voice message weight of each frame of original voice data;
s23, generating a voice message report according to the standard voice message weight of the original voice data of each frame.
In the invention, the weight operation is carried out on the voice message queue containing a plurality of voice signal characteristics, the voice message weight is filled into the blank report form according to the sequence from large to small, the position with larger weight indicates that the change of the audio signal is larger or more important, and the user can conveniently and rapidly obtain the important information of the voice data when viewing the voice message report form.
In the embodiment of the present invention, in S22, the voice message weight ρ of the n-th frame of original voice data n The calculation formula of (2) is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein r is n Voice message value r representing n-th frame of original voice data n+1 A voice message value representing the n+1th frame of original voice data, epsilon represents a minimum value, and N represents the total frame number of the original voice data.
In the embodiment of the present invention, in S24, the specific method for generating the voice message report is: and filling the standard voice message weight of each frame of original voice data into the voice message blank report according to the sequence from large to small to generate a voice message report.
Those of ordinary skill in the art will recognize that the embodiments described herein are for the purpose of aiding the reader in understanding the principles of the present invention and should be understood that the scope of the invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific modifications and combinations from the teachings of the present disclosure without departing from the spirit thereof, and such modifications and combinations remain within the scope of the present disclosure.

Claims (10)

1. A method for message-based data synchronization, comprising the steps of:
s1, acquiring original voice data, preprocessing the original voice data to generate de-noised voice data, inputting the original voice data and the de-noised voice data into a constructed voice processing model, and generating a voice message queue;
s2, determining voice message weight according to the voice message queue, and generating a voice message report;
s3, synchronously transmitting the voice message report and the denoising voice data to the user terminal.
2. The message-based data synchronization method according to claim 1, wherein S1 comprises the sub-steps of:
s11, collecting a plurality of frames of original voice data, denoising each frame of original voice data, and generating each frame of denoised voice data;
s12, constructing a voice processing model, inputting original voice data of each frame and denoising voice data of each frame into the voice processing model, and generating a voice message sequence;
s13, arranging all voice message values in the voice message sequence according to time sequence as a voice message queue.
3. The message-based data synchronization method according to claim 2, wherein in S12, the speech processing model includes a first feature extraction layer, a second feature extraction layer, an operator A1, a feature fusion layer, and a feature generation layer;
the input end of the first feature extraction layer and the input end of the second feature extraction layer are both used as input ends of a voice processing model; the output end of the first characteristic extraction layer is connected with the first input end of the arithmetic unit A1; the output end of the second characteristic extraction layer is connected with the second input end of the arithmetic unit A1; the output end of the arithmetic unit A1 is connected with the input end of the feature fusion layer; the output end of the characteristic fusion layer is connected with the input end of the characteristic generation layer; the output end of the characteristic generation layer is used as the output end of the voice processing model.
4. A message-based data synchronization method according to claim 3, wherein the expression of the first feature extraction layer is: x= { X 1 ,x 2 ,..,x n ,..,x N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, X represents the output of the first feature extraction layer, N represents the total number of frames of the original speech data, X n An audio feature value representing the original speech data of the nth frame, s representing the pitch of the original speech data, f representing the sampling frequency of the original speech data, l n Indicating the loudness of the original voice data of the nth frame, and e indicating an index;
the expression of the second feature extraction layer is: y= { Y 1 ,y 2 ,..,y m ,..,y M ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where m=1, 2, M, Y denote the output of the second feature extraction layer, M denote the total number of frames of the denoised speech data, Y m Audio characteristic value representing m-th frame of de-noised speech data,/->Pitch representing denoised speech data, l m Representing the loudness of the m-th frame of de-noised speech data.
5. A message-based data synchronization method according to claim 3, wherein the expression of operation A1 is: z= { x 1 -y 1 ,x 2 -y 2 ,..,x n -y m ,..,x N -y M N=m; where n=1, 2, N, m=1, 2, M, x n Audio feature value, y, representing n-th frame of original speech data m The audio characteristic value of the M-th frame of de-noised voice data is represented, N represents the total frame number of the original voice data, and M represents the total frame number of the de-noised voice data.
6. A message-based data synchronization method according to claim 3, wherein the expression of the feature fusion layer is:the method comprises the steps of carrying out a first treatment on the surface of the Wherein P represents the output of the feature fusion layer, σ k A weight parameter, b, representing the kth neuron in the feature fusion layer k The bias parameter of the kth neuron in the feature fusion layer is represented, Z represents the output of an operator A1, K represents the number of neurons in the feature fusion layer, max (DEG) represents the maximum value operation, X represents the output of the first feature extraction layer, and Y represents the output of the second feature extraction layer.
7. A message-based data synchronization method according to claim 3, wherein the expression of the feature generation layer is: r= { R 1 ,r 2 ,..,r n ,..,r N ｝，The method comprises the steps of carrying out a first treatment on the surface of the Where n=1, 2,..n, N represents the total number of frames of the original speech data, r n Voice message value, x, representing n-th frame of original voice data n An audio feature value representing the original speech data of the nth frame, and P representing the output of the feature generation layer.
8. The message-based data synchronization method according to claim 1, wherein S2 comprises the sub-steps of:
s21, acquiring a voice message blank report;
s22, calculating the voice message weight of each frame of original voice data according to the voice message queue to generate a voice message weight set;
s23, performing de-duplication processing on the voice message weight set to generate standard voice message weight of each frame of original voice data;
s23, generating a voice message report according to the standard voice message weight of the original voice data of each frame.
9. The method according to claim 8, wherein in S22, the voice message weight ρ of the n-th frame of original voice data is n The calculation formula of (2) is as follows:the method comprises the steps of carrying out a first treatment on the surface of the Wherein r is n Voice message value r representing n-th frame of original voice data n+1 A voice message value representing the n+1th frame of original voice data, epsilon represents a minimum value, and N represents the total frame number of the original voice data.
10. The message-based data synchronization method according to claim 8, wherein in S24, the specific method for generating the voice message report is: and filling the standard voice message weight of each frame of original voice data into the voice message blank report according to the sequence from large to small to generate a voice message report.
CN202311753182.5A 2023-12-20 2023-12-20 Data synchronization method based on message Active CN117440001B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311753182.5A CN117440001B (en) 2023-12-20 2023-12-20 Data synchronization method based on message

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311753182.5A CN117440001B (en) 2023-12-20 2023-12-20 Data synchronization method based on message

Publications (2)

Publication Number Publication Date
CN117440001A true CN117440001A (en) 2024-01-23
CN117440001B CN117440001B (en) 2024-02-27

Family

ID=89546565

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311753182.5A Active CN117440001B (en) 2023-12-20 2023-12-20 Data synchronization method based on message

Country Status (1)

Country Link
CN (1) CN117440001B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105489223A (en) * 2015-11-26 2016-04-13 成都微讯云通科技有限公司 Real-time voice transmission method
CN108388942A (en) * 2018-02-27 2018-08-10 四川云淞源科技有限公司 Information intelligent processing method based on big data
CN113782044A (en) * 2021-08-25 2021-12-10 慧言科技(天津)有限公司 Voice enhancement method and device
CN114299981A (en) * 2021-12-29 2022-04-08 中国电信股份有限公司 Audio processing method, device, storage medium and equipment
WO2023086311A1 (en) * 2021-11-09 2023-05-19 Dolby Laboratories Licensing Corporation Control of speech preservation in speech enhancement

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105489223A (en) * 2015-11-26 2016-04-13 成都微讯云通科技有限公司 Real-time voice transmission method
CN108388942A (en) * 2018-02-27 2018-08-10 四川云淞源科技有限公司 Information intelligent processing method based on big data
CN113782044A (en) * 2021-08-25 2021-12-10 慧言科技(天津)有限公司 Voice enhancement method and device
WO2023086311A1 (en) * 2021-11-09 2023-05-19 Dolby Laboratories Licensing Corporation Control of speech preservation in speech enhancement
CN114299981A (en) * 2021-12-29 2022-04-08 中国电信股份有限公司 Audio processing method, device, storage medium and equipment

Also Published As

Publication number Publication date
CN117440001B (en) 2024-02-27

Similar Documents

Publication Publication Date Title
CN105244042B (en) A kind of speech emotional interactive device and method based on finite-state automata
CN109256150A (en) Speech emotion recognition system and method based on machine learning
CN106409310A (en) Audio signal classification method and device
WO2022199215A1 (en) Crowd-information-fused speech emotion recognition method and system
CN111986661B (en) Deep neural network voice recognition method based on voice enhancement in complex environment
CN113129898B (en) Machine-assisted conference recording system and method
CN110070855A (en) A kind of speech recognition system and method based on migration neural network acoustic model
CN109063624A (en) Information processing method, system, electronic equipment and computer readable storage medium
CN110717410A (en) Voice emotion and facial expression bimodal recognition system
CN116665676B (en) Semantic recognition method for intelligent voice outbound system
CN117440001B (en) Data synchronization method based on message
CN115249479A (en) BRNN-based power grid dispatching complex speech recognition method, system and terminal
CN112420079B (en) Voice endpoint detection method and device, storage medium and electronic equipment
CN109545198A (en) A kind of Oral English Practice mother tongue degree judgment method based on convolutional neural networks
CN108010533A (en) The automatic identifying method and device of voice data code check
CN116611447A (en) Information extraction and semantic matching system and method based on deep learning method
CN115688868A (en) Model training method and computing device
CN107825433A (en) A kind of card machine people of children speech instruction identification
CN114550729A (en) Cry detection model training method and device, electronic equipment and storage medium
CN1062365C (en) A method of transmitting and receiving coded speech
CN113994427A (en) Source-specific separation of speech in audio recordings by predicting an isolated audio signal conditioned on speaker representation
CN212675912U (en) Automatic voice recognition system based on FPGA
CN110888986B (en) Information pushing method, device, electronic equipment and computer readable storage medium
CN117558264A (en) Dialect voice recognition training method and system based on self-knowledge distillation
US20220238095A1 (en) Text-to-speech dubbing system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant