Voice-to-text medical dialogue processing method and device, medium and equipment

Abstract

The invention provides a voice-to-text medical dialogue processing method and device, a medium and equipment, and the method comprises the steps: obtaining a voice file, sending the voice file to a voice large model for processing, converting the voice file into a preliminary file through the voice large model, adding a space at a proper position of the preliminary file, so as to divide the preliminary file into short sentence files, and storing the short sentence files. And adding punctuation marks in the short sentence file, identifying role information of each dialogue in the short sentence file, labeling an identity, converting the voice file into a text file, and inputting the text file into a text large model for performing structured processing to obtain a structured file. According to the method, recording errors caused by negligence, fatigue or understanding deviation of a hospital guide nurse are avoided, completeness and accuracy of medical dialogue information are guaranteed, medical terminologies can be accurately recognized and converted, understanding ambiguity caused by nonstandard terminologies is reduced, time for manually recording and sorting information is greatly saved, and working efficiency is improved. And personalized medical services are realized.

Claims

1. A speech-to-text medical dialogue processing method, characterized in that: The method comprises: Get the audio file; Sending the voice file to the voice big model for processing, the voice big model converts the voice file into a preliminary file, adds spaces at appropriate positions of the preliminary file to divide the preliminary file into short sentence files, adds punctuation marks to the short sentence files, identifies the role information of each dialogue in the short sentence files and marks the identity before converting the voice file into a text file; The text file is input into the text macro model for structural processing to obtain a structured document.

2. The method according to claim 1, characterized in that The method according to claim 1 is characterized in that the voice file is obtained in real time or by a preset time period.

3. The method according to claim 1, characterized in that After acquiring the voice file and before sending the voice file to the voice model for processing, the method further includes: Performing noise estimation on the speech file; Perform noise reduction processing on the speech file after noise estimation; Perform inverse transformation and post-processing on the denoised speech file to obtain an adjusted and reconstructed speech file.

4. The method according to claim 3, characterized in that After converting the voice file into a text file and before performing structured processing on the text file to obtain a structured document, the method further includes: Performing data preprocessing on the text file; The vocabulary in the data and processed text files is matched with the vocabulary in a pre-built professional vocabulary to standardize the vocabulary in the text files.

5. The method according to claim 4, characterized in that Matching the words in the text file after data preprocessing with the words in the pre-built professional vocabulary includes: Using string matching algorithms to match words in the pre-processed text files with words in a pre-built professional vocabulary; or Formulate vocabulary replacement rules according to the context, and match the vocabulary in the text file after data preprocessing with the vocabulary in the pre-built professional vocabulary according to the vocabulary replacement rules; or The words in the text file after data preprocessing are read into the data frame, and each line of data is looped through to match and replace the words therein, so that the words in the text file after data preprocessing are batch converted.

6. The method according to claim 1, characterized in that Inputting the text file into the text macro model for structural processing to obtain a structured document includes: Import patient history case information into the text model; An initial structured document instruction is received, and the text big model outputs an initial structured document corresponding to the initial structured document instruction from the text file according to the initial structured document instruction.

7. The method according to claim 6, characterized in that The method further comprises: Performing detection on the initial structured document; Adjusting the initial structured instruction according to the detection result to obtain an adjusted structured document instruction; The text big model outputs an adjusted structured document corresponding to the adjusted structured document instruction from the text file according to the adjusted structured document instruction; The steps of adjusting the structured document instructions and outputting the structured document according to the adjusted structured execution by the text large model are repeatedly performed until the structured document output by the text large model is accurately detected.

8. A speech-to-text medical dialogue processing device, characterized in that: The device comprises: A language file acquisition module is used to acquire voice files; A language-to-text module, used for sending the voice file to the voice big model for processing, the voice big model converts the voice file into a preliminary file, adds spaces at appropriate positions of the preliminary file to divide the preliminary file into short sentence files, adds punctuation marks to the short sentence files, identifies the role information of each dialogue in the short sentence files and marks the identity before converting the voice file into a text file; The structured document module is used to input the text file into the text macro model for structural processing to obtain a structured document.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to run the computer program to perform the method according to any one of claims 1 to 7.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 7.

Images