Laboratory research and development intelligent automatic data recording system based on speech recognition and NLP

Abstract

The invention discloses a laboratory research and development intelligent automatic data recording system based on voice recognition and NLP, and relates to the field of voice recognition. Comprising a voiceprint recognition module which is used for recognizing the voiceprint of a current researcher and abandoning voice signals of other people so as to achieve a noise removal function; the voice recognition module is used for recognizing the voice of the researcher and analyzing the voice content; the result output module is used for outputting the identification result; the data recording module is used for recording the output result; the recognition logic of the voiceprint recognition module is as follows: a voiceprint database is constructed, and a researcher inputs voiceprint information into the system in advance before experimental research; and voiceprint recognition: carrying out voiceprint feature extraction on the collected sound, and comparing the voiceprint feature with a database. According to the method, the voiceprint is recognized, so that the voice of people except for researchers can be effectively removed, a good'impurity removal 'effect is achieved, and the recording precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of speech recognition, and in particular to a laboratory-developed intelligent automatic data recording system based on speech recognition and NLP. Background Art

[0002] Data recording is an indispensable part of scientific research laboratories. Traditional manual recording methods are not only inefficient but also prone to errors. With the continuous development of artificial intelligence technology, especially the maturity of speech recognition (ASR) and natural language processing (NLP) technology, it is possible to realize intelligent and automatic recording of laboratory data. ASR technology can convert researchers' voice instructions into text information, while NLP technology can further parse these texts, understand their intentions, and automatically complete the corresponding data recording operations.

[0003] However, the existing technology cannot reliably recognize the researcher's voice, which causes "noise" to be generated when the researcher communicates with other people. This noise will affect the recognition results and cause the recording results to be less accurate.

[0004] To this end, the present invention proposes a laboratory-developed intelligent automatic data recording system based on speech recognition and NLP. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a laboratory-developed intelligent automatic data recording system based on speech recognition and NLP.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, including:

[0008] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0009] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0010] A result output module, which outputs the recognition result;

[0011] The data recording module records the output results.

[0012] Preferably: the recognition logic of the voiceprint recognition module is:

[0013] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0014] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0015] Preferably: the step A2 comprises the following steps:

[0016] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0017] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0018] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0019] Preferably: in step A21, the method for extracting voiceprint features comprises the following steps:

[0020] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0021] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0022] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0023] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0024] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0025] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0026] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0027] Preferably: in step A23, the spatial distance is:

[0028] Euclidean distance,

[0029] Mahalanobis distance,

[0030] Manhattan distance,

[0031] Chebyshev distance,

[0032] Cosine distance,

[0033] Any one of .

[0034] Preferably, the speech recognition module is composed of a plurality of speech recognition units.

[0035] Preferably: the working logic of the speech recognition module includes the following steps:

[0036] B1: Segment the speech after voiceprint recognition to obtain multiple segments of speech signals;

[0037] B2: Send each speech signal to a speech recognition unit to obtain multiple recognition results;

[0038] B3: Recombine the recognition results to obtain recognition semantics.

[0039] Preferably: the step B1 comprises the following steps:

[0040] B11: Based on the pre-established silence model, a deep learning algorithm is used to identify the language of each frame of the continuous speech signal to identify the silent language;

[0041] B12: Using silent speech as the segmentation point, the continuous speech signal is divided into multiple speech segments.

[0042] Preferably, the speech recognition module uses NLP technology to perform semantic analysis and intent recognition.

[0043] Preferably: the speech recognition module includes a lexical analyzer, a syntactic analyzer and a semantic understander, the lexical analyzer splits the text information into vocabulary units; the syntactic analyzer constructs the syntactic structure of the sentence; and the semantic understander extracts the user's intention based on the syntactic structure and lexical meaning.

[0044] The beneficial effects of the present invention are:

[0045] 1. The present invention can effectively remove the voices of people other than researchers by identifying voiceprints, thereby having a good "impurity removal" effect and increasing recording accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1This is a processing flow chart of a laboratory-developed intelligent automatic data recording system based on speech recognition and NLP proposed by the present invention. DETAILED DESCRIPTION

[0047] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0048] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", and "set" should be understood in a broad sense, for example, they can be fixedly connected or set, or detachably connected or set, or integrally connected or set. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0049] Embodiment 1:

[0050] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0051] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0052] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0053] A result output module, which outputs the recognition result;

[0054] The data recording module records the output results.

[0055] The recognition logic of the voiceprint recognition module is:

[0056] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0057] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0058] The step A2 comprises the following steps:

[0059] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0060] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0061] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0062] In step A21, the method for extracting voiceprint features includes the following steps:

[0063] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0064] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0065] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0066] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0067] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0068] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0069] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0070] In step A23, the spatial distance is: Euclidean distance,

[0071] Embodiment 2:

[0072] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0073] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0074] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0075] A result output module, which outputs the recognition result;

[0076] The data recording module records the output results.

[0077] The recognition logic of the voiceprint recognition module is:

[0078] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0079] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0080] The step A2 comprises the following steps:

[0081] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0082] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0083] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0084] In step A21, the method for extracting voiceprint features includes the following steps:

[0085] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0086] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0087] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0088] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0089] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0090] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0091] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0092] In step A23, the spatial distance is: Mahalanobis distance,

[0093] Embodiment 3:

[0094] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0095] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0096] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0097] A result output module, which outputs the recognition result;

[0098] The data recording module records the output results.

[0099] The recognition logic of the voiceprint recognition module is:

[0100] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0101] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0102] The step A2 comprises the following steps:

[0103] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0104] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0105] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0106] In step A21, the method for extracting voiceprint features includes the following steps:

[0107] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0108] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0109] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0110] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0111] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0112] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0113] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0114] In step A23, the spatial distance is Manhattan distance,

[0115] Embodiment 4:

[0116] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0117] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0118] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0119] A result output module, which outputs the recognition result;

[0120] The data recording module records the output results.

[0121] The recognition logic of the voiceprint recognition module is:

[0122] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0123] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0124] The step A2 comprises the following steps:

[0125] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0126] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0127] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0128] In step A21, the method for extracting voiceprint features includes the following steps:

[0129] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0130] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0131] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0132] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0133] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0134] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0135] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0136] In step A23, the spatial distance is: Chebyshev distance,

[0137] Embodiment 5:

[0138] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0139] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0140] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0141] A result output module, which outputs the recognition result;

[0142] The data recording module records the output results.

[0143] The recognition logic of the voiceprint recognition module is:

[0144] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0145] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0146] The step A2 comprises the following steps:

[0147] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0148] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0149] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0150] In step A21, the method for extracting voiceprint features includes the following steps:

[0151] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0152] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0153] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0154] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0155] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0156] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0157] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0158] In step A23, the spatial distance is: cosine distance,

[0159] Embodiment 6:

[0160] A laboratory-developed intelligent automatic data recording system based on speech recognition and NLP, comprising:

[0161] Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise;

[0162] The speech recognition module recognizes the researcher's speech and analyzes the speech content;

[0163] A result output module, which outputs the recognition result;

[0164] The data recording module records the output results.

[0165] The recognition logic of the voiceprint recognition module is:

[0166] A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research.

[0167] A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

[0168] The step A2 comprises the following steps:

[0169] A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features;

[0170] A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data;

[0171] A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

[0172] In step A21, the method for extracting voiceprint features includes the following steps:

[0173] A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part;

[0174] A212: Framing: Divide the pre-emphasized speech signal into short time frames;

[0175] A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion;

[0176] A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal;

[0177] A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale.

[0178] A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part;

[0179] A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

[0180] In step A23, the spatial distance is:

[0181] Euclidean distance,

[0182] Mahalanobis distance,

[0183] Manhattan distance,

[0184] Chebyshev distance,

[0185] Cosine distance,

[0186] Any one or more combinations of .

[0187] The speech recognition module is composed of a plurality of speech recognition units.

[0188] The working logic of the speech recognition module includes the following steps:

[0189] B1: Segment the speech after voiceprint recognition to obtain multiple segments of speech signals;

[0190] B2: Send each speech signal to a speech recognition unit to obtain multiple recognition results;

[0191] B3: Recombine the recognition results to obtain recognition semantics.

[0192] In the step B1, it comprises the following steps:

[0193] B11: Based on the pre-established silence model, a deep learning algorithm is used to identify the language of each frame of the continuous speech signal to identify the silent language;

[0194] B12: Using silent speech as the segmentation point, the continuous speech signal is divided into multiple speech segments.

[0195] The speech recognition module uses NLP technology to perform semantic analysis and intent recognition, and includes a lexical analyzer, a syntactic analyzer, and a semantic understander. The lexical analyzer splits text information into vocabulary units; the syntactic analyzer constructs the syntactic structure of the sentence; and the semantic understander extracts the user's intent based on the syntactic structure and lexical meaning.

[0196] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent automatic data recording system developed by a laboratory based on speech recognition and NLP, characterized in that: include: Voiceprint recognition module, which is used to identify the voiceprint of the current researcher and discard the voice signals of other personnel to achieve the function of removing noise; The speech recognition module recognizes the researcher's speech and analyzes the speech content; A result output module, which outputs the recognition result; The data recording module records the output results.

2. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 1, characterized in that: The recognition logic of the voiceprint recognition module is: A1: Construct a voiceprint database. Researchers enter voiceprint information into the system before conducting experimental research. A2: Voiceprint recognition: extract the voiceprint features of the collected sound and compare it with the database. If it does not match the corresponding researcher in the database, the sound signal will be discarded. If it matches, the next step of content recognition will be carried out.

3. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 2, characterized in that: The step A2 comprises the following steps: A21: After receiving the voice data, obtain the voiceprint features of the voice data, and construct a corresponding voiceprint feature vector based on the voiceprint features; A22: Input the voiceprint feature vector into the pre-trained background channel model to construct the current voiceprint identification vector corresponding to the voice data; A23: The spatial distance between the current voiceprint identification vector and the standard voiceprint identification vector corresponding to the voiceprint database is used to perform identity authentication based on the distance and generate a verification result.

4. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 3, characterized in that: In step A21, the method for extracting voiceprint features includes the following steps: A211: Pre-emphasis: Pre-process the input speech signal by applying a high-pass filter to enhance the energy of the high-frequency part and reduce the energy loss of the low-frequency part; A212: Framing: Divide the pre-emphasized speech signal into short time frames; A213: Windowing: Apply a window function to each frame to reduce spectral leakage and spectral distortion; A214: Fourier transform: Perform a fast Fourier transform (FFT) on each windowed frame to convert the time domain signal into a frequency domain signal; A215: Mel filter bank: Based on the human ear's perception of sound, a group of Mel filters are used to filter the frequency domain signal. The center frequencies of the filters are evenly distributed on the Mel frequency scale. A216: Logarithmic operation: Perform logarithmic operation on the output of each Mel filter to enhance the information of the low-frequency part; A217: Features: From the calculation results, the cepstral coefficients are used as the final feature vector.

5. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 3, characterized in that: In step A23, the spatial distance is: Euclidean distance, Mahalanobis distance, Manhattan distance, Chebyshev distance, Cosine distance, Any one of .

6. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 1, characterized in that: The speech recognition module is composed of a plurality of speech recognition units.

7. The laboratory research and development intelligent automatic data recording system based on speech recognition and NLP according to claim 6 is characterized in that: The working logic of the speech recognition module includes the following steps: B1: Segment the speech after voiceprint recognition to obtain multiple segments of speech signals; B2: Send each speech signal to a speech recognition unit to obtain multiple recognition results; B3: Recombine the recognition results to obtain recognition semantics.

8. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 7, characterized in that: In the step B1, it comprises the following steps: B11: Based on the pre-established silence model, a deep learning algorithm is used to identify the language of each frame of the continuous speech signal to identify the silent language; B12: Using silent speech as the segmentation point, the continuous speech signal is divided into multiple speech segments.

9. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 1, characterized in that: The speech recognition module uses NLP technology to perform semantic analysis and intent recognition.

10. The intelligent automatic data recording system for laboratory research and development based on speech recognition and NLP according to claim 9, characterized in that: The speech recognition module includes a lexical analyzer, a syntactic analyzer and a semantic understander. The lexical analyzer divides text information into vocabulary units; the syntactic analyzer constructs the syntactic structure of the sentence; and the semantic understander extracts the user's intention based on the syntactic structure and lexical meaning.

Images