Emergency rescue scene voice transmission method and device, electronic equipment and storage medium

Abstract

The invention provides an emergency rescue scene voice transmission method and device, electronic equipment and a storage medium, and the method comprises the steps: collecting original voice information, carrying out the preprocessing of the original voice information, and obtaining an audio signal; converting the audio signal into a short-time frame signal, and extracting a feature vector of each short-time frame signal; inputting the extracted feature vectors into an acoustic model, and identifying different phonemes contained in each feature vector; inputting the different phonemes into a language model, and predicting a phoneme sequence; and decoding the phoneme sequence, optimizing a decoding result in combination with emergency rescue scene information, outputting text information, and transmitting the text information to a preset terminal, thereby solving the problem that voice information cannot be transmitted due to limited message length and low bandwidth of Beidou short message transmission. The voice information is efficiently transmitted through the Beidou short message, and the accuracy of converting voice into characters is improved through the acoustic model and the language model.

Description

Technical Field

[0001] The present invention relates to the field of emergency rescue technology, and in particular to a method, device, electronic equipment and storage medium for voice transmission in an emergency rescue scenario. Background Art

[0002] In emergency communications, fast and clear voice information transmission is essential for making accurate decisions, conducting on-site assessments, and coordinating rescue efforts. In emergency rescue scenarios, Beidou short message communication is usually used. However, traditional Beidou short message communication does not support voice information transmission, but only supports text information transmission. In emergency rescue scenarios, distress messages can be sent faster and more efficiently through voice. Since Beidou short message communication cannot meet the transmission and analysis capabilities of voice information, the best rescue time is affected. Summary of the invention

[0003] The present invention provides a voice transmission method, device, electronic device and storage medium for emergency rescue scenarios, which are used to solve the defect that the communication means of traditional emergency rescue scenarios cannot meet the transmission and analysis capabilities of voice information, thus affecting the optimal rescue time.

[0004] The present invention provides a method for voice transmission in an emergency rescue scenario, comprising: Collecting original voice information, preprocessing the original voice information, and obtaining an audio signal; Convert the audio signal into a short-time frame signal, and extract a feature vector of each short-time frame signal; The extracted feature vectors are input into the acoustic model to identify the different phonemes contained in each feature vector; Inputting the different phonemes into a language model to predict a phoneme sequence; The phoneme sequence is decoded, the decoding result is optimized in combination with the emergency rescue scene information, text information is output, and the text information is transmitted to a preset terminal.

[0005] According to the emergency rescue scene voice transmission method provided by the present invention, the preprocessing of the original voice information includes: The original voice information is subjected to noise filtering processing.

[0006] According to the emergency rescue scene voice transmission method provided by the present invention, before the original voice information is preprocessed, it also includes: It is determined whether the memory size occupied by the original voice information is greater than a preset threshold, and if so, the original voice information is segmented.

[0007] According to the voice transmission method for emergency rescue scenarios provided by the present invention, the acoustic model is an acoustic model established based on state transition probability and observation probability, the state transition probability is the possibility of identifying another specific phoneme following a certain phoneme in a continuous speech stream, the observation probability is the possibility of hearing certain sound characteristics when a phoneme is given, and the acoustic model is used to identify different phonemes by combining the possibility of another specific phoneme following the certain phoneme and the possibility of hearing certain sound characteristics when a phoneme is given.

[0008] According to the emergency rescue scenario voice transmission method provided by the present invention, the acoustic model is an acoustic model trained based on a dialect database.

[0009] According to the emergency rescue scene voice transmission method provided by the present invention, the language model is an n-gram language model, and the different phonemes are input into the language model to predict the phoneme sequence, including: The n-gram language model is used to score multiple possible sequences composed of different phonemes, and the possible sequence with the highest score is selected as the phoneme sequence.

[0010] According to the emergency rescue scene voice transmission method provided by the present invention, the decoding of the phoneme sequence, optimizing the decoding result in combination with the emergency rescue scene information, and outputting text information include: Establish multiple word paths based on different phoneme sequences, and obtain the path with the highest cumulative probability among all possible paths as the most likely phoneme sequence; When predicting the most likely phoneme sequence pair of words, it is first compared with the relevant vocabulary of the emergency rescue scene to obtain the optimal word as text information.

[0011] The present invention also provides a voice transmission device for emergency rescue scenarios, comprising: The acquisition module is used to acquire original voice information, pre-process the original voice information, and obtain an audio signal; A conversion module, used for converting the audio signal into a short-time frame signal and extracting a feature vector of each short-time frame signal; A first input module is used to input the extracted feature vectors into the acoustic model to identify different phonemes contained in each feature vector; A second input module, used for inputting the different phonemes into a language model to predict a phoneme sequence; The output module is used to decode the phoneme sequence, optimize the decoding result in combination with the emergency rescue scene information, output text information, and transmit the text information to a preset terminal.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the emergency rescue scenario voice transmission method as described in any one of the above items is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the emergency rescue scenario voice transmission method described in any one of the above items is implemented.

[0014] The method, device, electronic device and storage medium for voice transmission in emergency rescue scenarios provided by the present invention collect original voice information, pre-process the original voice information and obtain an audio signal; convert the audio signal into a short-time frame signal and extract a feature vector of each short-time frame signal; input the extracted feature vector into an acoustic model to identify different phonemes contained in each feature vector; input the different phonemes into a language model to predict a phoneme sequence; decode the phoneme sequence, optimize the decoding result in combination with the emergency rescue scenario information, output text information, and transmit the text information to a preset terminal, thereby solving the problem that voice information cannot be transmitted due to the limited message length of Beidou short message transmission and low bandwidth, realizing efficient transmission of voice information through Beidou short messages, and improving the accuracy of voice-to-text conversion through acoustic models and language models. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 This is one of the flow charts of the method for voice transmission in an emergency rescue scenario provided by an embodiment of the present invention; Figure 2 This is a second flow chart of a method for voice transmission in an emergency rescue scenario provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the structure of a voice transmission system provided by an embodiment of the present invention; Figure 4 It is a functional structure diagram of a voice transmission device for emergency rescue scenarios provided by an embodiment of the present invention; Figure 5 It is a functional structure diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Figure 1 A flowchart of a method for voice transmission in an emergency rescue scenario provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the emergency rescue scene voice transmission method provided by the embodiment of the present invention includes: Step 101: Collect original voice information, pre-process the original voice information, and obtain an audio signal; In the embodiment of the present invention, using voice to call for help in an emergency rescue scenario is a more timely way of calling for help. Therefore, collecting audio information sent by the user through a terminal device is more suitable for emergency rescue scenarios.

[0019] Step 102: convert the audio signal into a short-time frame signal, and extract a feature vector of each short-time frame signal; Since the audio signal is a continuous speech signal, in order to improve the timeliness and accuracy of feature extraction, the continuous signal can be first converted into a short-time frame signal.

[0020] Step 103: input the extracted feature vectors into the acoustic model to identify different phonemes contained in each feature vector; In an embodiment of the present invention, the acoustic model is, for example, a model based on deep learning, such as a convolutional neural network, a recurrent neural network, or a long short-term memory network. The task of the acoustic model is to predict possible phonemes based on these feature vectors. For example, the word "Hello" can be predicted as the following phoneme sequence: / h / / ə / / l / / oʊ / . The acoustic model will make a judgment on each time frame and give a probability distribution of which phoneme the frame is most likely to represent. In this way, the entire audio will be parsed into a series of continuous phonemes.

[0021] Step 104: input the different phonemes into a language model to predict a phoneme sequence; In an embodiment of the present invention, multiple phonemes are further combined to form words. In this process, a language model is used to improve recognition accuracy, because some phoneme sequences may correspond to multiple possible words. The language model helps select the most reasonable phoneme sequence based on context information.

[0022] Step 105: decode the phoneme sequence, optimize the decoding result in combination with the emergency rescue scenario information, output text information, and transmit the text information to a preset terminal.

[0023] In the embodiment of the present invention, by adding a dialect database to the model training and combining emergency scenarios for key judgment in decoding, the accuracy of recognition is improved and the communication efficiency of the Beidou terminal is greatly improved.

[0024] In emergency rescue scenarios, Beidou short message communication is usually used. However, traditional Beidou short message communication does not support the transmission of voice information, but only supports the transmission of text information. In emergency rescue scenarios, distress messages can be sent faster and more efficiently through voice. Since Beidou short message communication cannot meet the transmission and analysis capabilities of voice information, the best time for rescue is affected.

[0025] The embodiment of the present invention provides a voice transmission method for emergency rescue scenarios, which collects original voice information, pre-processes the original voice information, and obtains an audio signal; converts the audio signal into a short-time frame signal, and extracts a feature vector of each short-time frame signal; inputs the extracted feature vector into an acoustic model to identify different phonemes contained in each feature vector; inputs the different phonemes into a language model to predict a phoneme sequence; decodes the phoneme sequence, optimizes the decoding result in combination with the emergency rescue scenario information, outputs text information, and transmits the text information to a preset terminal, thereby solving the problem that voice information cannot be transmitted due to the limited message length of Beidou short message transmission and low bandwidth, and realizing efficient transmission of voice information through Beidou short messages, and improving the accuracy of voice-to-text conversion through acoustic models and language models.

[0026] Based on any of the above embodiments, Figure 2 As shown, the preprocessing of the original voice information includes: It is determined whether the memory size occupied by the original voice information is greater than a preset threshold. If so, the original voice information is segmented and each segment of the voice information is subjected to noise filtering.

[0027] If not, noise filtering is performed on the original voice information.

[0028] Accordingly, after receiving the information, if the cloud control platform determines that the voice information has been segmented on the computing platform, it will perform data splicing, otherwise it will directly display the text information.

[0029] Based on any of the above embodiments, the acoustic model is an acoustic model established based on state transition probability and observation probability, wherein the state transition probability is the possibility of identifying another specific phoneme following a certain phoneme in a continuous speech stream, and the observation probability is the possibility of hearing certain sound characteristics when a phoneme is given. The acoustic model is used to identify different phonemes by combining the possibility of another specific phoneme following the certain phoneme and the possibility of hearing certain sound characteristics when a phoneme is given.

[0030] In some embodiments of the present invention, the acoustic model is an acoustic model trained based on a dialect database.

[0031] Based on any of the above embodiments, the language model is an n-gram language model, and the step of inputting the different phonemes into the language model to predict a phoneme sequence includes: The n-gram language model is used to score multiple possible sequences composed of different phonemes, and the possible sequence with the highest score is selected as the phoneme sequence.

[0032] In the embodiment of the present invention, decoding the phoneme sequence, optimizing the decoding result in combination with the emergency rescue scene information, and outputting text information includes: Establish multiple word paths based on different phoneme sequences, and obtain the path with the highest cumulative probability among all possible paths as the most likely phoneme sequence; When predicting the most likely phoneme sequence pair of words, it is first compared with the relevant vocabulary of the emergency rescue scene to obtain the optimal word as text information.

[0033] During speech recognition, the acoustic model first converts the audio signal into a series of possible phoneme sequences. The n-gram language model then scores these sequences and selects those that are the most linguistically reasonable and likely. In this way, the final text output not only reflects the information of the audio signal, but is also linguistically optimized to be more coherent and in line with actual language habits. Suppose there is a sentence: "I want to go to the rescue location." In a 3-gram model, when processing the word "to", the model will look at the previous two words "I want" and predict the probability of "to" based on the frequency of the "I want to" triple in the training data. If the next word is "the", the model will further check the "want to the" triple and predict the probability of "the" based on this. This process helps reduce recognition errors and makes the transcription of the entire sentence more accurate.

[0034] like Figure 3As shown, the voice transmission system includes a front-end APP, a computing power platform, a Beidou terminal, and an operation control platform. The emergency rescue scene voice transmission method provided by the embodiment of the present invention includes the following steps: The front-end APP has the functions of voice collection and forwarding. Users use the front-end APP to send voice emergency information. The front-end APP collects the audio information sent by the user and forwards it to the computing power platform. After receiving the collected audio information, the computing platform first determines the memory size of the collected audio. If it is larger than the threshold, it will be segmented. Otherwise, it will be directly filtered for noise. Perform feature extraction on the preprocessed audio signal, convert the audio signal into short-time frames, and extract the feature vector of each frame; By training the database with dialects, the state transition probability and observation probability are established to form an acoustic model, and the acoustic model is used to identify different phonemes and words from the speech features of the audio information.

[0035] The n-gram language model is used to improve recognition accuracy, and the next word is predicted based on historical context (n-1 words) and sentence structure to form the most likely phoneme sequence.

[0036] In the decoding stage, the acoustic model and language model are combined to search for the optimal phoneme sequence through a dynamic programming algorithm. Combined with the emergency rescue site, key information such as location and danger is judged and text information is output; After the computing platform completes the conversion of voice information into text information, it sends the text information to the cloud control platform through the Beidou terminal and short message; After the cloud control platform receives the information, if the information has been segmented on the computing power platform, data splicing will be performed, otherwise the information will be displayed directly.

[0037] The voice transmission method for emergency rescue scenarios provided by the embodiment of the present invention can ensure that voice information is output using the Beidou intelligent terminal and transmitted to the operation control platform in real time. When performing Beidou short message communication, the voice information is compressed into text information and then sent, thereby expanding the problem that Beidou short message communication cannot perform real-time data transmission after the collected voice information. By adding a dialect database to the model training and combining the emergency scenario for key judgment in decoding, the accuracy of recognition is improved, and the communication efficiency of the Beidou terminal is greatly improved.

[0038] The emergency rescue scene voice transmission device provided by the present invention is described below. The emergency rescue scene voice transmission device described below and the emergency rescue scene voice transmission method described above can be referenced to each other.

[0039] Figure 4A schematic diagram of the structure of a voice transmission device for emergency rescue scenarios provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown, the emergency rescue scene voice transmission device provided by the embodiment of the present invention includes: The acquisition module 401 is used to acquire original voice information, pre-process the original voice information, and obtain an audio signal; A conversion module 402, configured to convert the audio signal into a short-time frame signal and extract a feature vector of each short-time frame signal; The first input module 403 is used to input the extracted feature vectors into the acoustic model to identify different phonemes contained in each feature vector; A second input module 404, configured to input the different phonemes into a language model to predict a phoneme sequence; The output module 405 is used to decode the phoneme sequence, optimize the decoding result in combination with the emergency rescue scene information, output text information, and transmit the text information to a preset terminal.

[0040] The emergency rescue scene voice transmission device provided by the embodiment of the present invention collects original voice information, pre-processes the original voice information, and obtains an audio signal; converts the audio signal into a short-time frame signal, and extracts a feature vector of each short-time frame signal; inputs the extracted feature vector into an acoustic model to identify different phonemes contained in each feature vector; inputs the different phonemes into a language model to predict a phoneme sequence; decodes the phoneme sequence, optimizes the decoding result in combination with the emergency rescue scene information, outputs text information, and transmits the text information to a preset terminal, thereby solving the problem that voice information cannot be transmitted due to the limited message length of Beidou short message transmission and low bandwidth, and realizing efficient transmission of voice information through Beidou short messages, and improving the accuracy of voice-to-text conversion through acoustic models and language models.

[0041] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The memory 530 includes a computer program, an operating system and acquired data, and the processor 510 may call the logic instructions in the memory 530 to execute the emergency rescue scene voice transmission method, the method comprising: collecting original voice information, pre-processing the original voice information, and acquiring an audio signal; converting the audio signal into a short-time frame signal, extracting a feature vector of each short-time frame signal; inputting the extracted feature vector into an acoustic model, identifying different phonemes contained in each feature vector; inputting the different phonemes into a language model, and predicting a phoneme sequence; decoding the phoneme sequence, optimizing the decoding result in combination with the emergency rescue scene information, outputting text information, and transmitting the text information to a preset terminal.

[0042] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the relevant technology or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0043] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it is implemented to execute the emergency rescue scene voice transmission method provided by the above-mentioned methods. The method includes: collecting original voice information, preprocessing the original voice information, and obtaining an audio signal; converting the audio signal into a short-time frame signal, and extracting a feature vector of each short-time frame signal; inputting the extracted feature vector into an acoustic model to identify different phonemes contained in each feature vector; inputting the different phonemes into a language model to predict a phoneme sequence; decoding the phoneme sequence, optimizing the decoding result in combination with the emergency rescue scene information, outputting text information, and transmitting the text information to a preset terminal.

[0044] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0045] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A voice transmission method for emergency rescue scenarios, characterized in that: include: Collecting original voice information, preprocessing the original voice information, and obtaining an audio signal; Convert the audio signal into a short-time frame signal, and extract a feature vector of each short-time frame signal; The extracted feature vectors are input into the acoustic model to identify the different phonemes contained in each feature vector; Inputting the different phonemes into a language model to predict a phoneme sequence; The phoneme sequence is decoded, the decoding result is optimized in combination with the emergency rescue scene information, text information is output, and the text information is transmitted to a preset terminal.

2. The method for voice transmission in emergency rescue scenarios according to claim 1, characterized in that: The preprocessing of the original voice information comprises: The original voice information is subjected to noise filtering processing.

3. The method for voice transmission in emergency rescue scenarios according to claim 1 or 2, characterized in that: Before the original voice information is preprocessed, the method further includes: It is determined whether the memory size occupied by the original voice information is greater than a preset threshold, and if so, the original voice information is segmented.

4. The method for voice transmission in emergency rescue scenarios according to claim 1, characterized in that: The acoustic model is an acoustic model established based on state transition probability and observation probability, wherein the state transition probability is the probability of recognizing another specific phoneme immediately following a certain phoneme in a continuous speech stream, and the observation probability is the probability of hearing certain sound characteristics when a phoneme is given; The acoustic model is also used to identify different phonemes by combining the likelihood of another specific phoneme following the certain phoneme and the likelihood of hearing certain sound characteristics when a given phoneme is heard.

5. The method for voice transmission in emergency rescue scenarios according to claim 1 or 4, characterized in that: The acoustic model is an acoustic model trained based on a dialect database.

6. The method for voice transmission in emergency rescue scenarios according to claim 1, characterized in that: The language model is an n-gram language model, and the different phonemes are input into the language model to predict a phoneme sequence, including: The n-gram language model is used to score multiple possible sequences composed of different phonemes, and the possible sequence with the highest score is selected as the phoneme sequence.

7. The method for voice transmission in emergency rescue scenarios according to claim 1, characterized in that: The decoding of the phoneme sequence, optimizing the decoding result in combination with the emergency rescue scene information, and outputting text information includes: Establish multiple word paths based on different phoneme sequences, and obtain the word path with the highest cumulative probability among all possible word paths as the most likely phoneme sequence; When predicting the most likely phoneme sequence pair of words, it is first compared with the relevant vocabulary of the emergency rescue scene to obtain the optimal word as text information.

8. A voice transmission device for emergency rescue scenarios, characterized in that: include: The acquisition module is used to acquire original voice information, pre-process the original voice information, and obtain an audio signal; A conversion module, used for converting the audio signal into a short-time frame signal and extracting a feature vector of each short-time frame signal; A first input module is used to input the extracted feature vectors into the acoustic model to identify different phonemes contained in each feature vector; A second input module, used for inputting the different phonemes into a language model to predict a phoneme sequence; The output module is used to decode the phoneme sequence, optimize the decoding result in combination with the emergency rescue scene information, output text information, and transmit the text information to a preset terminal.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the emergency rescue scene voice transmission method as described in any one of claims 1 to 7 is implemented.

10. A non-transitory readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the emergency rescue scene voice transmission method as described in any one of claims 1 to 7 is implemented.

Images