Audio analog signal recovery method based on deep learning
By introducing digital semantic information in traditional FM communication and using deep learning models to encode and decoding audio simulation signals, the traditional audio simulation signal communication is susceptible to noise interference, and high -quality audio simulation signals are realized.
Patent Information
- Application Number
- CN202411657378.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-19
- Publication Date
- 2025-01-28
AI Technical Summary
Traditional audio analog signal communication is susceptible to noise interference, resulting in a decrease in signal quality and making it difficult to achieve high-fidelity transmission, especially when hardware resources are limited.
By introducing digital semantic information in traditional FM communication, the deep learning model is used to encode and decoding the audio simulation signal, generate and use semantic information auxiliary signals to recover to improve the quality of audio simulation communication.
The quality of audio simulation signals under the guarantee of compatibility and low -bandwidth transmission conditions improves the quality of the audio simulation signal, improves the effect of de -no -noise and distorted correction, and the quality of the original signal is close to the quality of the original signal.
Smart Images
Figure FT_1 
Figure FT_2
Abstract
Description
Technical Field
[0001] The present invention relates to the field of wireless communication technology, and in particular to a method and system for restoring an audio analog signal using deep learning technology. Background Art
[0002] Traditional audio analog signal communications are susceptible to noise interference, resulting in signal quality degradation and difficulty in achieving high-fidelity transmission.
[0003] Although digital communication technology has made significant progress, analog signals are still widely used in radio, television, walkie-talkies and other fields due to their advantages such as low cost, low energy consumption, no cliff effect and a large user base.
[0004] How to improve the quality of audio analog signals, especially when hardware resources are limited, has become an urgent problem to be solved. Summary of the invention
[0005] In view of this, the present invention proposes an audio analog signal recovery system and method based on deep learning.
[0006] Specifically, the technical solution of this application is to construct an audio analog signal recovery model by introducing digital semantic information into traditional FM communication, aiming to improve the quality of audio analog communication while ensuring compatibility with existing equipment. It includes:
[0007] Step 1: At the transmitter, the original audio signal is transmitted through an analog FM channel, and the converted semantic information is transmitted through a low-bandwidth digital communication layer.
[0008] Step 2: At the receiving end, the received semantic information is used to assist in the recovery of the received analog signal to improve the audio performance.
[0009] Step 3: Use the deep learning model to encode the original audio signal at the sender to generate semantic information.
[0010] Step 4: Use the deep learning model to decode the analog signal at the receiving end and restore the audio based on the received semantic information. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1 It is a flowchart of a sound simulation signal recovery system based on deep learning shown in an exemplary embodiment of the present application.
[0012] Figure 2 It is a detailed structural diagram of a sound simulation signal recovery system based on deep learning shown in an exemplary embodiment of the present application. DETAILED DESCRIPTION
[0013] The following examples illustrate the present invention, but the present invention is not limited to these examples. Modifications to the specific embodiments of the present invention or equivalent replacement of some technical features without departing from the spirit of the present invention should be included in the scope of the technical solution claimed by the present invention.
[0014] The terms used in one or more embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present application. The singular forms of "a", "said" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any possible combination of one or more associated listed items.
[0015] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".
[0016] See also Figure 1 , Figure 1 A flowchart of a deep learning-based analog signal recovery system provided according to an embodiment of the present application is shown.
[0017] The audio analog signal sample data is input into a feature conversion network of a pre-trained analog signal restoration system to obtain a semantic encoding of the audio analog signal sample.
[0018] In the example of the present application, in order to simplify the operation of restoring the audio analog signal and improve the efficiency of restoring the audio analog signal, end-to-end audio analog signal restoration can be achieved through a network model that integrates semantic information feature extraction and restoration.
[0019] The semantic encoder at the transmitting end uses a deep learning model to encode the original audio signal to generate semantic information. The analog signal encoder at the receiving end receives and decodes the semantic information, and uses the residual vector quantization module to compress the semantic information for transmission on the digital communication layer.
[0020] The digital communication layer uses LoRa technology, which can achieve low-bandwidth semantic information transmission while ensuring coverage. LoRa technology was chosen based on its low power consumption and low bandwidth consumption, and is suitable for analog signal recovery systems that need to run long-term in remote monitoring scenarios.
[0021] In this embodiment, the decoder at the receiving end combines the received semantic information with the received analog signal to perform audio recovery to improve audio performance. The design of the decoder takes into account the denoising and distortion correction of the audio signal to ensure that the quality of the restored audio is as close to the original signal as possible.
[0022] See also Figure 2 In one embodiment of the present application, the cascaded deep semantic information extraction network and restoration network are trained in the following manner.
[0023] For any training sample in the training set, it is input into the deep feature extraction network to obtain the deep semantic information features of the training sample.
[0024] In an embodiment of the present application, before restoring the audio analog signal through the cascaded deep semantic information extraction network and restoration model, it is necessary to train the above-mentioned network and restoration network using a training set including a certain number of training samples until the network converges, and then perform the audio analog signal restoration task.
[0025] The semantic encoding features of the training sample are input into the restoration network to obtain the restoration result of the training sample.
Claims
1. A method for restoring an audio analog signal based on deep learning, characterized in that: The following steps are involved: At the transmitting end, the original audio signal is transmitted through the analog FM channel, while the converted semantic information is transmitted through the low-bandwidth digital communication layer; At the receiving end, the received semantic information is used to assist in the recovery of the received analog signal to improve the audio performance; Use deep learning models to encode the original audio signal at the sender to generate semantic information; The deep learning model is used to decode the analog signal at the receiving end and perform audio recovery in combination with the received semantic information.
2. The method for restoring an analog audio signal according to claim 1, characterized in that: The low-bandwidth digital communication layer adopts LoRa technology, which can achieve low-bandwidth semantic information transmission while ensuring coverage.
3. The method for restoring an analog audio signal according to claim 1, characterized in that: The deep learning model includes a semantic encoder, a residual vector quantization module and a decoder, which are used to extract semantic features of the original audio signal and compress semantic information at the transmitting end, receive and decode the semantic information at the receiving end, and perform audio recovery.
4. The method for restoring an analog audio signal according to claim 1, characterized in that: The semantic information includes a feature vector of the original audio signal, and the feature vector is compressed by a residual vector quantization module to facilitate transmission on a digital communication layer.
5. The method for restoring an analog audio signal according to claim 1, characterized in that: The audio restoration process includes combining the received semantic information with the received analog signal to improve the audio performance, including denoising and distortion correction of the signal.
6. An audio analog signal recovery system, characterized in that: include: A transmitting end, comprising a semantic encoder for extracting semantic features of the original audio signal, and a residual vector quantization module for compressing the semantic information; A receiving end, including a vector quantization encoder for receiving and decoding semantic information, and a decoder for audio recovery; Digital communication layer, used to transmit semantic information between sender and receiver.
7. The audio analog signal recovery system according to claim 6, characterized in that: The digital communication layer is a LoRa communication module, which is used to transmit semantic information between a transmitting end and a receiving end.
8. The analog signal recovery system according to claim 6, characterized in that: The system also includes a hardware implementation part, including an FM transmitter, a LoRa transceiver, a computer, etc., which is used to realize communication between the sending end and the receiving end.