Method and communication system for improving the playback quality of audio signals

Abstract

The invention relates to a method and a communication system for improving the playback quality of audio signals which have been transmitted by radio, in particular by an aircraft (10), as a radiotelephony message (11), wherein k) at least one unchanged audio signal (1) is received by a communication system (12), the unchanged audio signal comprising a useful signal and an interfering signal, and the interfering signal being composed of a plurality of interfering signal components, l) the interfering signal components are removed from the audio signal, and an air traffic controller workstation (13) connected to the communication system plays said audio signal back to an air traffic controller (14) in real time as an audio signal which has been filtered by a filter, m) a differential signal between the unchanged audio signal and the filtered audio signal is formed in parallel with the playback at the air traffic controller workstation, n) a device for quality inspection checks the differential signal for any voice components of the radiotelephony message which may be present, and o) if a predetermined quality threshold value is not reached, an action at the air traffic controller workstation is triggered.

Description

[0001] Method and communication system for improving the playback quality of audio signals

[0002] The invention relates to a method for improving the playback quality of audio signals transmitted via radio, in particular from an aircraft, as a radiotelephony message. The invention also relates to a communication system for receiving unaltered audio signals transmitted via radio, in particular from an aircraft, as a radiotelephony message, comprising a computer system that includes: a filter having an input for receiving an unaltered audio signal, and an output for transmitting a filtered audio signal to an audio device at an air traffic controller's workstation.

[0003] Radio communication is regularly subject to signal interference, in which the actual audio signal, consisting of sequences of human speech, is overlaid with noise from various sources, some of which are technically uncontrollable. Analog radio, used in air traffic control and monitoring, as well as in maritime communication, for example, significantly limits the audio quality (bandwidth approximately 300 Hz to 3 kHz) of the transmitted speech signals. Other interference, generated by the environment or the transmission itself, can also affect communication. Depending on the severity, this interference can lead to reduced speech intelligibility or even complete unintelligibility. Therefore, it is desirable to improve the playback quality of received audio signals using technical means.

[0004] AT 526211 Al describes such a procedure, in which an air traffic controller can decide during radio communication whether to have a section of a radio message, which they have just heard but find too distorted, played back in a filtered version. For this purpose, filtered audio segments are created by an electronic system in parallel with the unaltered radio transmission and kept ready for playback. A disadvantage of this method is that, in addition to their regular duties, the air traffic controller must also decide whether to have sections of the radio message repeated. Furthermore, in some situations, it is unpredictable whether the playback of the filtered section will have significantly better sound quality. It is also unpredictable whether sections will be filtered too heavily and essential parts of the speech will be suppressed.

[0005] The aim of the invention is to create a method that improves the quality of radiotelephony messages in real time while meeting safety requirements, for example, in aviation communications. Furthermore, it should provide air traffic controllers with the ability to adapt the system to their specific audio quality preferences.

[0006] The invention for achieving these goals is a method in which

[0007] a) at least one unaltered audio signal is received from a communication system, wherein the unaltered audio signal comprises a useful signal and a noise signal, the noise signal being formed by several noise signal components, b) the audio signal is reduced by noise signal components and is reproduced in real time as a filtered audio signal to an air traffic controller via an air traffic controller's workstation connected to the communication system, wherein c) a difference signal between the unaltered audio signal and the filtered audio signal is generated in parallel with the reproduction at the air traffic controller's workstation, d) a quality control device checks the difference signal for any voice components of the radiotelephony message that may be present, and

[0008] e) an action is triggered at the air traffic controller's workstation if a predetermined quality threshold is not met.

[0009] One embodiment of the method involves generating at least one indicator during the quality check in step d), based on at least one numerical value representing the quantity of any voice components present in the difference signal. A quality threshold is defined for this indicator in step e). This allows the filter to be automatically switched off and, if necessary, back on. Furthermore, the indicator values ​​can be displayed to the user to provide additional information for better assessing the filtering quality. Another embodiment of the method involves reducing the type and / or parameters of the filter or playing back the unaltered audio signal instead of the filtered audio signal at the air traffic controller's workstation. This also makes it particularly possible to adapt the filtering effect to varying levels of interference.

[0010] One embodiment of the method involves terminating the action after a predetermined period of time, or terminating the action when the quality threshold is exceeded again for a predetermined period. The advantage of this approach is that it relieves the user of the need to adjust the filter through automation, thereby allowing them to concentrate on the actual voice message.

[0011] One implementation of the procedure involves the air traffic controller manually ending the action. This option is advantageous for users who wish to retain some control over the system.

[0012] One embodiment of the method involves additionally transmitting an acoustic or visual signal to the air traffic controller during the action. The visual signal, in particular, either indicates that the filter is inactive or displays a scale value for the filter quality. Such additional information enhances the user's understanding in a specific radio communication situation characterized by varying levels of interference.

[0013] One embodiment of the method involves recording the unaltered audio signal, with particular emphasis on marking sections of the recorded audio signal at the points in time when an action began or, if applicable, ended. Advantageously, this allows an audio segment to be listened to again for comparison purposes in particularly problematic situations. This improves the understanding of the voice message.

[0014] One embodiment of the method involves playing back a previous segment of the unaltered audio signal of a predetermined length. This advantageously allows the user to assess the degree of distortion in the unfiltered signal.

[0015] One embodiment of the method involves recording the filtered audio signal. Advantageously, this recording can be used for evaluation and training purposes after the communication process.

[0016] One embodiment of the method involves using at least one unaltered audio signal as an analog audio signal, particularly one transmitted via VHF voice radio. This makes the method applicable to an important frequency range in aeronautical radio communications.

[0017] One embodiment of the method involves using a digital filter. Digital filters are more effective and flexible than analog filters.

[0018] One embodiment of the method involves selecting and / or modifying the type and / or parameters of the filter at the air traffic controller's workstation, particularly by the controller. This increases the controllability and, if necessary, the adaptability of the air traffic controller's workstation to changing users with different hearing preferences.

[0019] One embodiment of the method involves selecting and / or modifying the type and / or parameters of the filter based on the quality check. Advantageously, this allows the method to be dynamically adapted to changing reception conditions.

[0020] One embodiment of the method involves creating at least one file, in particular a text file, a spreadsheet, or a database, which includes at least: the recorded unaltered audio signal, the recorded filtered audio signal, the type and parameters of the filter, the times of the actions, and a radio channel identification. This allows for an evaluation of the method results after several radio events to improve the method for future radio events. The aforementioned communication system achieves its objectives by further including a quality control device in the computer system, which is in data communication with the filter and, moreover, with a graphical user interface (GUI) at the air traffic controller's workstation.

[0021] One embodiment involves the quality control device generating at least one indicator during the quality control process. This indicator comprises at least one numerical value representing the quantity of any voice components present in the difference signal, and a quality threshold is predefined for the indicator. This allows the filter activity to be automatically switched off and, if necessary, back on. Furthermore, it enables the indicator values ​​to be displayed to the user, providing additional information to better understand a current radio message.

[0022] One embodiment results from the computer system being programmed and / or configured to execute a method according to the steps described above.

[0023] The invention is explained in more detail below with reference to exemplary embodiments illustrated in the drawings. Figure 1 shows the schematic path of a radio signal from an aircraft to the air traffic controller's workstation, and Figure 2 shows a block diagram of the processing of an audio signal.

[0024] The proposed solution concerns a method for quality assurance in the signal enhancement of speech signals during safety-critical and noise-prone communication (e.g., Air Traffic Management, ATM, radio communication between pilots and air traffic controllers, or in the maritime sector, e.g., Global Maritime Distress and Safety System, GMDSS). This method is applied in real time and ensures that no speech information is filtered and thus lost.

[0025] Direct intervention in live communication, specifically altering a received speech signal before or during its output to the receiver's headphones or speakers (such as those of an air traffic controller), is particularly problematic if it cannot be guaranteed that significant speech components are filtered out or suppressed during the processing of the original signal. If audio signals are altered, it must be ensured that this occurs without any loss of information.

[0026] As shown in Fig. 1, a radiotelephony message 11 is transmitted by an aircraft 10 and received by a ground station 15, as usual. The audio signal 1 is processed by a communication system 12 and played back at an air traffic controller's workstation 13. In addition to the output device 4 (speaker, headphones, etc.), the air traffic controller 14 can be provided with additional data on the radio situation visually via a display and input device 8 (“Graphical User Interface”, GUI), such as a screen and a computer keyboard.

[0027] Part of the communication system 12 is shown in Fig. 2. The audio signal 1 is electronically filtered by a filter 2 in an audio improvement service module and passed on as a filtered signal 3 to the output device 4. Simultaneously, a differential signal 6 is generated by the filter 2 and passed to a detector used by the quality control system 5.

[0028] Evaluating the difference signal 6 (original minus processed signal) enables quality assurance of the audio enhancement, as it is checked for any remaining speech. This ensures that no speech information is lost. This can be done in real time or with stored signals. If the audio processing Z-filtering works optimally, the difference signal 6 contains only background noise and no speech fragments. For evaluating the difference signal 6, the modulation index, which is commonly used in audio file analysis, could be employed, for example.

[0029] The audio signal 1 is therefore simultaneously

[0030] a) output to the user (e.g. air traffic controller 14) via loudspeaker or headset (output device 8), optionally

[0031] al the incoming unprocessed audio signal 1, or

[0032] a.2 the improved speech signal (filtered audio signal 3); b) and the evaluation of the difference signal 6 and an indication (using known so-called Key Performance Indices, KPIs 7) for the user as to whether relevant information has been filtered out (quality assurance), whereby

[0033] c) The user is additionally provided with a time-limited audio segment, defined by the detector in quality check 5, either the unprocessed audio signal (unfiltered) or again through a filter (but with higher latency for improved audio quality), for playback.

[0034] The user can select the original audio signal 1 or the enhanced, filtered audio signal 3 for playback. This function is implemented in the so-called "VCS" (Voice Communication System), the (speech) switching system to which the radio receivers are connected. A special feature is the ability to use the described quality assurance function to automatically deactivate the active sound enhancement algorithm (e.g., upon identification of speech components in the differential signal 6), thus preventing any loss of information.

[0035] In a further step, the selection and settings of the audio algorithms of filter 2 can also be automated and dependent on the evaluation of the difference signal 6. This can further optimize speech quality and make the work of the air traffic controller 14 even more efficient.

[0036] The calculation of the audio algorithms, i.e., the improvement applied to the speech signal, is carried out using a service as a component of the VCS software and runs in real time during the playback of the recorded speech signal or is calculated directly on the live audio stream.

[0037] To improve speech quality, various filter types can be used as filter 2. In the case of digital filters, in addition to computational effort (and memory requirements), the latency—that is, the delay of the audio signal caused by signal processing—must also be considered. A parametric equalizer (center frequency, amplitude, and filter quality are adjustable) allows specific frequencies or frequency ranges to be boosted or cut. This can reduce narrowband and static background noise without significantly affecting the intelligibility of the speech signal. However, such filters are less suitable for reducing broadband background noise that lies essentially within or overlaps with the speech frequency range.

[0038] When using analog VHF two-way radio, speech is typically overlaid with broadband noise. To filter this broadband noise sufficiently, more complex algorithms are employed. These can be classic adaptive filters (e.g., Wiener filters) or AI-based filters (e.g., Convolutional Neural Networks, CNNs, or Recurrent Neural Networks, RNNs). In the case of AI-based filters, a very high reduction (> 20 dB) of noise can be achieved through appropriate training on typical background noise without affecting the speech signal. The quality of such AI-based filters, i.e., the maximum possible reduction of noise without affecting the speech, depends on the computational effort (and memory requirements) as well as the latency and is usually significantly higher than that of classic filters.

[0039] To meet the requirements of common Air Traffic Control standards (ATC standards, e.g., EUROCAE ED-136) regarding the overall latency (delay of the transmitted audio signal) of an IP-based voice communication system, filter 2 should introduce only minimal latency (approximately 10 ms). However, a higher latency (e.g., approximately 50 or 100 ms) allows for better identification of interference and the voice signal, and thus also better separation of these signal components. If, due to the requirements specified in the ATC domain, the latency in the direct signal path must be minimal, the evaluation of the differential signal 6 with the quality control detector 5 can generate additional information at a higher latency (e.g., 50 or 100 ms). This additional information ensures that no voice information is lost in safety-critical communication.The higher latency of the detector is still sufficiently low to automatically deactivate Filter 2 and / or trigger a warning for the user. Alternatively, Air Traffic Controller 14 can listen to an unsatisfactory audio segment processed with Filter 2 again using the "Short Term Recording" function, i.e., short-term voice recording, and activate another filter (e.g., a filter as known from AT 526211 A1, a prior art document) to improve speech intelligibility. The higher latency of the filter used here does not have a negative impact because Air Traffic Controller 14 anticipates a certain amount of time for listening to the radio transmission again in this situation anyway. Furthermore, Air Traffic Controller 14 considers the ongoing communication to be briefly interruptible and actually does interrupt it to listen to the critical audio segment again.

[0040] Filter 2, located directly in the signal path, can be switched on and off by the user or its effect adjusted (degree of noise reduction). The detector acting on the differential signal is always active when Filter 2 is switched on and provides a continuously updated indicator, e.g., KPIs 7, for the quality of the signal filtering, i.e., the proportion of speech fragments. The detector thus provides a continuous value that is checked against a predefined threshold. The resulting indicator can be displayed to air traffic controller 14, e.g., as a numerical value, in categories (e.g., A, B, C, or similar), as a color code, or as a moving pointer on a scale.

[0041] According to Fig. 2, the KPIs 7 can be used in two ways: firstly, as a (mainly graphical) display at the air traffic controller's workstation 13, to provide the air traffic controller 14 with visual information about the quality of the filtering activity during communication; and secondly, to be fed back to the filter 2, either for the automatic shutdown of the filter 2 if the filter quality is too low, or for ongoing adjustments to the filter parameters in filter 2.

[0042] It is also possible not only to make or change different settings of filter 2, but also to select the filter type / algorithm in the direct signal path. The evaluation of different filter types (or improved AI models) of filter 2 is facilitated by the stored (comparative) data provided by the second filter. Especially in the area of ​​AI-based filters, there is continuous development that may necessitate (recurring) evaluation. The following KPIs or metrics can be used for the qualitative assessment of the signal filtering – these can be used for both the incoming audio signal 1 and the difference signal 6:

[0043] • Signal level, i.e., amplitude of the audio signal,

[0044] • Crest factor, i.e., ratio of peak value to RMS value,

[0045] • Signal-to-noise ratio, i.e., the ratio of useful signal power to noise power (filtered signal to difference signal),

[0046] • Frequency spectrum, frequency components averaged over a time window - here a distinction could be made between typical and atypical patterns,

[0047] • Quality index (applicable only to the difference signal 6).

[0048] With filter 2 activated, the signal level of the differential signal 6 provides a direct indication of the level of interference and whether it is within a range typical for aviation radio communications. The signal-to-noise ratio of filtered signal 3 to differential signal 6 also allows for an assessment of the incoming signal quality.

[0049] Crucial for quality assurance is the calculation of the quality index based on the analysis of the difference signal (6). In the case of a KL-based second filter within the quality check (5), the model would, for example, deliver values ​​between 0 and 1, where 1 would be a pure useful signal (speech) and 0 a pure interference signal (noise). The value ranges or thresholds can be determined by analyzing recorded data (e.g., legal recordings).

[0050] The procedure can be easily integrated into a GUI. Firstly, GUIs that continuously provide real-time information about the radio channel are already mandatory at air traffic controller workstations; secondly, the KPIs to be displayed consist of numerical values ​​that can be conveniently arranged in the vicinity of other already displayed numerical values, or of symbols and color indicators (e.g., green background for numbers indicating optimal filtering), which can be just as easily incorporated into existing graphical display procedures.

Claims

Patent claims 1. Method for improving the playback quality of audio signals transmitted via radio, in particular from an aircraft (10), as a radiotelephony message (11), wherein f) at least one unaltered audio signal (1) is received from a communication system (12), wherein the unaltered audio signal (1) has a useful signal and a noise signal, the noise signal being formed by several noise signal components, g) the audio signal (1) is reduced by noise signal components and is reproduced in real time via an air traffic controller workstation (13) connected to the communication system (12) to an air traffic controller (14) as an audio signal (3) filtered by a filter (2), h) a difference signal (6) is generated between the unaltered audio signal (1) and the filtered audio signal (3), in parallel with playback at the air traffic controller's workstation (13), i) a quality control device (5) checks the difference signal (6) for any voice components of the radiotelephony message (11) that may be present, and j) an action is triggered at the air traffic controller's workstation (13) if a predetermined quality threshold is not met.

2. Method according to claim 1, characterized in that in step d) during the quality check (5) at least one indicator (7) is formed which is based on at least one numerical value which represents the quantity of any voting shares present in the difference signal (6) and wherein a quality threshold value for step e) is predetermined for the indicator (7).

3. Method according to claim 1 or 2, characterized in that the action consists of reducing the type and / or parameters of the filter (2) or reproducing the unaltered audio signal (1) instead of the filtered audio signal (3) at the air traffic controller's workstation (13).

4. Method according to claim 3, characterized in that the action is terminated after a predetermined period of time, or to terminate the action when the quality threshold is exceeded again for a predetermined period of time.

5. Method according to claim 3, characterized in that the action is terminated manually by the air traffic controller (14).

6. Method according to one of claims 1 to 5 characterized in that the action additionally comprises the transmission of an acoustic or optical signal to the air traffic controller (14), wherein the optical signal in particular either indicates that the filter is not active or indicates a scale value for the filter quality.

7. Method according to one of claims 1 to 6, characterized in that the unaltered audio signal (1) is recorded, wherein in particular sections of the recorded audio signal (1) are marked at the times when an action has started or possibly ended.

8. Method according to claim 7, characterized in that the action comprises replaying a past section of the unaltered audio signal (1) of a predetermined length.

9. Method according to one of claims 1 to 8, characterized in that the filtered audio signal (3) is recorded.

10. Method according to one of claims 1 to 9, characterized in that the at least one unaltered audio signal (1) is an analog audio signal (1), in particular transmitted via VHF voice radio.

11. Method according to any one of claims 1 to 10, characterized in that the filter (2) is a digital filter.

12. Method according to one of claims 1 to 11, characterized in that the type and / or parameters of the filter (2) at the air traffic controller's workstation (13), in particular by the air traffic controller (14), are selected and / or can be changed.

13. Method according to one of claims 1 to 12, characterized in that the type and / or parameters of the filter (2) are selected and / or can be changed based on the quality check (5).

14. Method according to one of claims 1 to 13, characterized in that at least one file, in particular a text file, a table file or a database, is created, which at least comprises: the recorded unaltered audio signal (1), the recorded filtered audio signal (3), type and parameters of the filter (2), times of the actions, and a radio channel identification.

15. Communication system (12) for receiving unaltered audio signals (1) transmitted by radio, in particular by an aircraft (10) as a radiotelephony message (11), comprising a computer system, wherein the computer system includes: a filter (2) which has an input for receiving an unaltered audio signal (1) and an output for transmitting a filtered audio signal (3) to an audio device (4) of an air traffic controller's workstation (13), characterized in that the computer system further comprises a quality control device (5) which is in data communication with the filter (2), wherein the device is furthermore in data communication with a screen display (8) of the air traffic controller's workstation (13).

16. Communication system (12) according to claim 15, characterized in that in the device for quality control (5) during the quality control (5) at least one indicator (7) can be formed which includes at least one numerical value which represents the quantity of any existing vote shares in the difference signal (6) and wherein a quality threshold value is predetermined for the indicator (7).

17. Communication system (12) according to claim 15 or 16, characterized in that the computer system is programmed and / or configured to execute a method according to one of claims 1 to 9.

Images