Method for determining a characteristic of a timepiece movement

Abstract

Method for determining a characteristic, in particular a chronometric characteristic, of a timepiece movement (100) or of a timepiece (200), comprising: - a step of acquiring a signal emitted by the timepiece movement (100) or the timepiece (200), - a step of decomposing the acquired signal so as to obtain a decomposed signal by using a transformation into discrete wavelets expressed in a base, in particular into Daubechies or Symlet type wavelets, - a step of filtering the decomposed signal so as to obtain a filtered signal, and - a step of analyzing the filtered signal so as to obtain the characteristic.

Description

[0001] Method for determining a characteristic of a watch movement.

[0002] The invention relates to a method for determining a characteristic of a watch movement or a timepiece. The invention also relates to a diagnostic or authentication method implementing such a determination method. The invention also relates to a determination device implementing the determination method or the diagnostic or authentication method. The invention further relates to a program product enabling the implementation of the determination method or the diagnostic or authentication method. The invention further relates to a recording medium enabling the implementation of the determination method or the diagnostic or authentication method. Finally, the invention relates to a signal from a data carrier.

[0003] Document EP2872950B1 describes a method for authenticating a watch by applying a controlled shock to generate a vibration, which is then analyzed and compared to reference signals stored in a database. Among the various signal analysis techniques that can be used, the document mentions, in particular, the Gabor transform, the Wigner transform, the Fourier transform, and the use of Morlet functions.

[0004] Documents EP2753986B1 and EP2872951 B1 describe methods using vibrations emitted by a watch during its operation to authenticate it. The analysis is performed in the frequency domain, for example, by wavelet decomposition. Wavelet transform denoising and wavelet decomposition are mentioned.

[0005] Document W02015025049A1 describes a method for authenticating a watch by measuring and analyzing the acoustic vibrations emitted by a quartz resonator within the watch. The aim of the invention is to improve existing methods for determining characteristics. In particular, the invention proposes a simple and reproducible method for determining the precise characteristics of watches, especially chronometric characteristics.

[0006] According to the invention, a determination method is defined by claim 1.

[0007] Methods of carrying out the determination process are defined by claims 2 to 14.

[0008] According to the invention, a first diagnostic or authentication method is defined by claim 15.

[0009] According to the invention, a second diagnostic or authentication method is defined by claim 16.

[0010] Execution methods for the diagnostic or authentication process are defined by claim 17.

[0011] According to the invention, a determination device is defined by claim 18.

[0012] According to the invention, a program product is defined by claim 19.

[0013] According to the invention, a recording medium is defined by claim 20.

[0014] According to the invention, a signal from a data carrier is defined by claim 21.

[0015] The accompanying drawings illustrate, by way of example, one embodiment of a determination method according to the invention. Figure 1 is a graph illustrating the temporal evolution of an acoustic signal emitted by the escapement of a watch movement over a period of a few hundredths of a second.

[0016] Figure 2 is a graph illustrating the time evolutions of a filtered signal obtained after decomposition and then filtering steps according to the invention of the acoustic signal of Figure 1.

[0017] Figure 3 illustrates an example of a wavelet that can be used to decompose a signal.

[0018] Figure 4 illustrates a family of Daubechies wavelets (db5) particularly suited to decompose an acoustic signal produced by the operation of an escapement in a clock movement.

[0019] Figure 5 is a graph illustrating the time evolution of an acoustic signal emitted by an escapement of a watch movement over a period of approximately 5 seconds.

[0020] Figure 6 illustrates a five-level decomposition of the signal in Figure 5 using Daubechies5 (db5) wavelets.

[0021] Figures 7 to 11 illustrate in more detail the decomposition of the signal from Figure 5 at each level. Figure 7 illustrates the decomposition at level 1. Figure 8 illustrates the decomposition at level 2. Figure 9 illustrates the decomposition at level 3. Figure 10 illustrates the decomposition at level 4. Figure 11 illustrates the decomposition at level 5. Figure 12 is a graph illustrating the temporal evolution of an acoustic signal emitted by the escapement of a watch movement over a period of approximately 5 seconds.

[0022] Figure 13 is a graph illustrating the time evolutions of the signal in Figure 12 after decomposition and then filtering according to the invention.

[0023] Figure 14 is another graph illustrating the time evolution of an acoustic signal emitted by an escapement of a watch movement over a period of a few hundredths of a second.

[0024] Figure 15 is a graph illustrating the time evolution of the signal from Figure 14 after decomposition and filtering according to the invention. It also illustrates a subsequent step in detecting the onset of exhaust shocks.

[0025] Figure 16 is a schematic view of an embodiment of a determination device according to the invention.

[0026] The inventors' work has established that frequency analysis of acoustic measurements of watch movements and watch parts allows:

[0027] - to carry out chronometric measurements, and / or

[0028] - to characterize a watch movement 100 or a timepiece 200 by establishing a signature of this watch movement and this timepiece.

[0029] The inventors determined, in particular, that the wavelet transform, applied especially to acoustic signals from piezoelectric sensors in measuring equipment, is very powerful for highlighting the natural frequencies of a clockwork system with high resolution, as well as for eliminating various sources of noise and spurious signals. The inventors' work demonstrated that wavelet transforms can detect the beginning and end of the various shocks of the escapement very robustly, despite the presence of different surrounding noise sources. The wavelet transform method allows, in particular:

[0030] - the acoustic measurement of the actual amplitude, and

[0031] - the detection of intermediate shocks of the exhaust between release and fall.

[0032] This makes it possible to make chronometric measurements more reliable and to create an expert system (a decision-making tool based on measurements and known rules).

[0033] Figures 1 and 2 illustrate an example of extracting the signature of a caliber with filtering of spurious signals. Figure 1 shows a first raw acoustic signal obtained by measurement. Figure 2 shows a second signal obtained after filtering the first signal according to the invention.

[0034] The primary source of sound in a mechanical watch is the escapement of the watch movement during its operation. Chronometric measurements (rate, amplitude, and timing mark) are based on detecting the acoustic signal generated by the various shocks of the escapement, as described below:

[0035] - A first noise is produced when the pallet fork touches the anchor fork entry, then the anchor is driven to disengage the pallet from the escape wheel. The impact of the pallet on the fork constitutes a first shock (shock 1).

[0036] A second noise occurs when a tooth of the escape wheel falls onto an impulse plane of the pallet fork and, as a result, the anchor fork catches the pallet fork pin. The impact of the escape wheel tooth on the impulse plane constitutes a second shock (shock 2). The impact of the fork on the pin constitutes a third shock (shock 3).

[0037] The third noise is generally the loudest. It occurs when a tooth of the escape wheel falls onto the pallet's resting surface (fall) and the anchor rod strikes a limiting pin (of the anchor). The impact of the escape wheel tooth on the resting surface constitutes a fourth noise (noise 4). The impact of the anchor rod against a limiting pin (stopper) constitutes a fifth noise (noise 5).

[0038] In an acoustic measurement according to the prior art, the first noise (shock 1) can be used to calculate the rate and the reference point. The second noise (shocks 2 and 3) is "very irregular and is not used for evaluation" [according to the training document "Measurement Techniques and Analysis of Watch Defects" published by Witschi, the company that manufactures the most widely used chronometric measuring equipment]. The third noise (shocks 4 and 5) is used to calculate the amplitude. Known prior art devices cannot:

[0039] - nor distinguish between shocks 2 and 3 in the second noise,

[0040] - nor distinguish between shocks 4 and 5 in the third noise.

[0041] According to the invention, the wavelet transform is used. Indeed, the wavelet transform allows for high resolution in both the time and frequency domains. The signal is first analyzed with a large scale to analyze the major features, and then the signal is analyzed with smaller scales to extract the finer features. The wavelet transform uses wavelet functions, or wavelets with different scales. This is illustrated in Figure 3. A wavelet is localized in time and allows for good temporal resolution in addition to frequency information.

[0042] The acoustic measurements of the watch movement or timepiece's operation are therefore analyzed using wavelet transformations. More precisely, according to the invention, the signal is transformed into discrete wavelet transforms (DWT) or continuous wavelet transforms (CWT). The continuous wavelet transform is just as relevant as the discrete wavelet transform but requires more computing resources. In summary, the principle of decomposition with DWT is to decompose the signal into two distinct components using high-pass and low-pass filters. The filter outputs produce, respectively, the detail coefficients cD (from the high-pass filter) and the approximation coefficients cA (from the low-pass filter).The decomposition can then be repeated to increase frequency resolution, notably by subjecting the approximation coefficients to decomposition using high-pass and low-pass filters before downsampling them. This analysis structure is called a filter bank. The functions used for the high-pass and low-pass filters can be, for example, the Daubechies (db5) or Symlets (sym5) wavelet families, which have proven to be particularly well-suited to the acoustic signals of watch movements and allow the capture of the finest characteristics of these signals. Daubechies wavelets are illustrated in Figure 4.

[0043] The detailed filtering process described below is based on a wavelet transform, which allows for the precise and robust extraction of shock instants of the escapement (release, impulse and fall), making it particularly suitable for chronometry, diagnostic and authentication applications.

[0044] As an example, Figure 5 illustrates the acoustic signal of a wristwatch measured using a sensor, which is visibly disturbed by background noise.

[0045] The Daubechies5 (db5) wavelet family is deployed to perform a five-level decomposition. Figure 6 illustrates the coefficients obtained by implementing a cascade of filters.

[0046] The acoustic signal is decomposed using cA approximation coefficients (low-pass, left in Figure 6) across 5 decomposition levels with the Daubechies5 function (db5). Similarly, the acoustic signal is decomposed using cD detail coefficients (high-pass, right in Figure 6) across 5 decomposition levels with the Daubechies5 function (db5).

[0047] Figures 7 through 11 illustrate the signal decomposition in more detail. In each of Figures 7 through 11, the original acoustic signal is shown in the upper right. Furthermore, in each of Figures 7 through 11, signals constructed from the coefficients cA and cD obtained by filtering (according to the different levels) or Daubechies wavelet decomposition of the original acoustic signal (according to the different levels) are shown to the right, below the representation of the original acoustic signal. The frequency distribution of each signal shown on the right is illustrated by a graph to the left of each signal. This frequency distribution is determined using the Fourier transform.

[0048] Levels 1 and 2 (in Figures 7 and 8) allow for the extraction of high frequencies and correspond to the components of the acoustic signal that best characterize the ticking sound of the escapement. They contain the most important characteristics of the acoustic measurement for defining the chronometric characteristics of the watch movement or timepiece.

[0049] Level 3 (in Figure 9) allows the extraction of the component of the sound which corresponds to the disturbance of the acoustic signal produced by the watch movement or the timepiece.

[0050] Level 4 (in Figure 10) allows us to extract mainly the components related to sound damping after shocks.

[0051] Level 5 (in Figure 11) primarily extracts the sound components related to the intrinsic noise of the watch movement or timepiece (i.e., the noise produced between ticks and clicks, when there is no shock). The decompositions allow for a detailed analysis of the acoustic signal to identify a correspondence with physical events, that is, to identify the moments when physical events occur, in particular to identify the moments when shocks from the escapement occur.

[0052] Algorithms were developed based on this decomposition to achieve effective denoising and better distinguish the different shocks during the escapement's operation. A filtering process was implemented to obtain improved resolution on the filtered signals. This filtering method considers neighboring coefficients (cD) within a given window to optimally adapt a filtering threshold according to the noise level. The filtering results are illustrated in Figures 12 and 13. The filtering eliminates parasitic noise (external noise sources, measurement noise). The different shocks of the escapement are thus discriminated very distinctly.

[0053] Figures 1 and 2 illustrate a "tic" (a complete function of the escapement), and the filtering eliminates high-frequency oscillations. With this filtering, the low-frequency background noise (parasitic signal) between the ticks and clicks is completely suppressed. This clearly highlights the moment the release begins (when the platter pin touches the anchor fork inlet) as well as the other shock moments. The end of the shocks is also clearly indicated, opening up new possibilities. For example, detecting the end of the fall eliminates the need to consider the angle between shocks and provides a unique definition of the amplitude determined by acoustic measurement.

[0054] The filtered signal can be further processed by applying a threshold to the signal envelope, for example, calculated using a Hilbert transform. This results in a signal that closely resembles a step function, thus enabling the determination of shock instants with high precision and reducing computation time. Figures 14 and 15 illustrate the shock start and end detection processing applied to the filtered signal and the resulting output. Figure 14 shows the raw acoustic signal. Figure 15 shows the filtered signal and the result of the processing performed on the filtered signal. The filtering and subsequent processing reliably and explicitly highlight all shocks from the exhaust and detect their start and end using a simple and robust algorithm, unaffected by background noise. Shock start times are represented by dashed lines in Figure 15.In comparison, current methodologies used in commercial equipment do not allow for pulse detection (shocks 2 and 3 previously mentioned). Pulse detection allows for better resolution and understanding of the measurement performed, robust calculation of timing parameters, and diagnostics. Figure 15 shows the dates, including the following:

[0055] - shock 1 (t1): 3.9496 s,

[0056] - shock 4 (t4): 3.9570 s (start of the fall).

[0057] We can therefore deduce the following duration:

[0058] - At_14=7.4 ms between dates t1 and t4.

[0059] For example, the duration At_14 and / or the dates t1 and t4 allow for a very precise calculation of the amplitude of the pendulum's oscillations. The determination of the amplitude of the pendulum's oscillations can be performed without optical measurement.

[0060] For example, the time between several shocks of the same type allows us to determine the rate of the watch movement or the timepiece.

[0061] For example, the difference in duration between several shocks of the same type during successive alternations makes it possible to determine the reference point of the watch movement or timepiece. In all these situations, the very precise determination of the shock times using the method that is the subject of the invention allows for an improved and very precise determination of chronometric parameters, in particular the chronometric parameters mentioned above.

[0062] Thus, an analysis step may include detecting the shock moments of an escapement, particularly the intermediate shocks between the impulse and the fall, including the start and / or end times of shocks, and using these moments to determine a timing characteristic. These moments can be determined with high precision.

[0063] More generally than what has been described above, in one embodiment of the method according to the invention, a characteristic, in particular a chronometric characteristic, of the watch movement 100 or the timepiece 200 is determined. To this end, the method comprises:

[0064] - a step of acquiring a signal emitted by the watch movement 100 or by the timepiece 200,

[0065] - a step of decomposing the acquired signal in order to obtain a decomposed signal using a transformation into discrete wavelets expressed in a basis, in particular into wavelets in an orthonormal basis,

[0066] - a filtering step of the decomposed signal to obtain a filtered signal,

[0067] - a step of analyzing the filtered signal in order to obtain the characteristic.

[0068] To do this, a device 1 for determining the characteristic is used. An embodiment of the determination device is described below with reference to Figure 16. The determination device comprises:

[0069] - a determination module 2,

[0070] - a sensor 3. Sensor 3 is, for example, a microphone adapted to acquire the sound signal emitted by the watch movement 100 and / or by the timepiece 200. Alternatively, sensor 3 can be an accelerometer. In such a case, the accelerometer can be fixed to a frame 190 of the watch movement 100 or to a component of a case of the timepiece 200, in particular to a case middle 103.

[0071] The determination module 2 includes:

[0072] - a logical processing unit 21,

[0073] - a memory 22, and

[0074] - a human-machine interface 23 or a machine-machine interface.

[0075] The logical processing unit 21 can be a microprocessor enabling the implementation of the decomposition, filtering, and analysis steps mentioned above. To this end, the microprocessor includes computing means for executing an algorithm or a program product for implementing the process that is the subject of the invention.

[0076] Memory 22 is capable of storing the algorithm or program product mentioned above. It also allows for the storage of intermediate or final results obtained during the implementation of the process that is the subject of the invention.

[0077] The human-machine interface may include a keyboard 231 and a screen 232, allowing a user to exchange data with the determination device. In the case of a machine-machine interface, the determination device 1 may exchange data with any other system, including an artificial intelligence system for the diagnosis and / or authentication of watch movements and / or watch parts. Alternatively, the artificial intelligence system may be integrated into the determination device.

[0078] The process has been described as applied to a standard Swiss lever escapement. However, the process is also applicable to any type of escapement (such as a detent escapement, a Robin escapement, a bidirectional tangential impulse escapement, a sequential distribution escapement, etc.).

[0079] The method has been described as applied to the determination of chronometric parameters by acoustic measurement of exhaust noise. However, the method is also applicable to a variety of measurements, including:

[0080] - acoustic and / or vibrational and / or laser vibrometric measurements following controlled or random impulse shocks;

[0081] - a determination of chronometric parameters by optical or opto-acoustic measurement;

[0082] - a measurement and listening to the noises in operation of the movement (escapement 110, gear train, automatic winding system 130, manual winding system, drive system 140 of a jump display, etc.) and / or of the watch (noises of the movement transmitted and modified by the case 103, rotation and click of a rotating bezel 102, opening / closing of a clasp 101, etc.).

[0083] The method has been described as applied to the signal produced by the escapement 110 of the watch movement 100. However, the method can be applied to the signal emitted by a regulating organ 120 during its operation or by an automatic winding system 130 or by a manual winding system or by a date drive system 140 or by a chronograph 150 or by a clasp 101 or by a rotating bezel 102.

[0084] The signal can be emitted during a current phase of operation of the timepiece or during a specific phase of operation of the timepiece such as a winding, a display jump, including a date jump, or a correction of a display.

[0085] Studies conducted by the inventors show that decomposed and filtered signals allow for a better identification of signatures specific to different types of calibers. Indeed, the filtered signal, or parameters of the filtered signal, can constitute a vibrational or acoustic signature of a watch movement or a timepiece. In particular, the filtered signal can be used to determine a component signature of a watch movement or a timepiece.

[0086] The method has been described as applied to acoustic chronometric measurements. It could alternatively or complementarily be applied to optical chronometric measurements and improve the accuracy of chronometric measurements.

[0087] The method according to the invention can be applied to other types of signals, such as optical signals as well as signals from clothing (rotating glasses and clasps).

[0088] The filtered signal according to the invention can be used to train an artificial intelligence system for the diagnosis and / or authentication of watch movements and / or watch parts. The artificial intelligence system may include an artificial neural network system.

[0089] The invention also relates to a diagnostic or authentication method, comprising:

[0090] - a step in implementing the determination process described above applied to a single 100-hour watch movement, and / or

[0091] - a casing step of the watch movement 100 and a step of implementing the determination method according to one of the preceding claims applied to the cased watch movement 100.

[0092] The invention also relates to a diagnostic or authentication method, comprising:

[0093] - a step in implementing the determination process described above, applied to a watch movement 100 or a watch part 200,

[0094] - a step of applying a controlled mechanical stress to the watch movement 100 or to the watch part 200, in particular by a controlled element for applying the controlled mechanical stress,

[0095] - a step of acquiring and analyzing the signal emitted by the watch movement 100 or by the watch part 200 in response to the controlled mechanical stress.

[0096] In such a diagnostic or authentication process, the machine-to-machine interface of the determination module 2 can control the application element of the controlled mechanical stress. The controlled mechanical stress can cause a shock to the watch movement 100 or the watch part 200, or a vibration of the watch movement 100 or the watch part 200, or an acceleration of the watch movement 100 or the watch part 200.

[0097] In these diagnostic or authentication processes, a training phase can be implemented. This training phase may include the implementation of the steps in either of the two preceding paragraphs and the processing of the analysis results using machine learning, for example, a neural network.

[0098] More generally, the filtered signal can be used, in the analysis step, to perform a diagnosis or authentication of a component of a watch movement 100 or a component of a timepiece 200 or of a watch movement 100 or a timepiece 200. This analysis step may include a comparison with reference and / or previous measurements on the same watch movement 100 or on the same timepiece 200 or on the same model of watch movement 100 or on the same model of timepiece 200. Complementarily or alternatively, the filtered signal can, in the analysis step, be used to identify the natural frequencies of the blanks 190 of a watch movement 100 or of a case 103 of a timepiece 200.

[0099] Complementarily or alternatively, the filtered signal can, in the analysis stage, be used to identify natural frequencies of components of a watch movement or of components of a watch part.

[0100] The process according to the invention offers numerous advantages, including:

[0101] - Effective suppression of various noise sources, with automatic adaptation to varying noise levels, thus optimizing its efficiency,

[0102] - a purification (or processing) of the acoustic signal in order to achieve a robust extraction of chronometric parameters (step, amplitude and reference point),

[0103] - an extraction of intermediate shocks between the release and the fall, allowing for a more precise diagnostic analysis.

[0104] - an extraction of a pure signature of the watch movement and a quantification of the influence of the case of the timepiece or the watch case on the sound of the ticking of the escapement.

[0105] - High-resolution frequency analysis of the acoustic signal allows for the characterization of sound propagation in the different materials of the case, particularly the case middle, thus facilitating diagnosis and authentication. Therefore, the filtered signal obtained in the process according to the invention can, in the analysis step, be used to determine the nature of the materials of components of a watch movement 100 or a watch part 200.

[0106] Analyzing the "tick-tock" sound produced by the escapement of a timepiece, particularly a watch, is essential in watchmaking for measuring, analyzing, and regulating the movement within the timepiece. Specifically, it allows for the analysis of the watch movement's rate and the amplitude of the oscillator's oscillations, particularly those of the balance wheel. However, prior to the invention described in this document, the analysis of the acoustic signal emitted by watches suffered from severe limitations. Thanks to this invention, the natural frequencies of the systems can be identified with high resolution. Furthermore, the invention enables extremely reliable denoising to eliminate spurious signals and the intrinsic noise of the measuring equipment: it is thus possible to distinguish the impulse phase from the release and fall phases of the watch movement's escapement.This is not achievable with known state-of-the-art techniques.

[0107] Beyond its application to chronometric measurements (rate, amplitude, and reference point), the invention can be applied to the diagnosis of a watch movement or timepiece (by differentiating a problematic measurement from a normal measurement). Thus, the invention can be used to identify defective components of a watch movement. Furthermore, the invention can be used to authenticate a watch movement or timepiece, particularly by extracting or determining its signature.

[0108] As previously discussed, wavelet decomposition is valuable for processing chronometric measurements, particularly for measuring the rate, amplitude, and reference point of a mechanical watch. As mentioned, the method enables the precise detection of escapement shock moments for extracting chronometric parameters and for diagnostics: it is notably possible to detect the end of the escapement's function and extract the actual amplitude. This allows for an amplitude estimate obtained via an acoustic method that is as accurate as an amplitude estimate obtained via an optical method, which was not the case prior to the invention. As also explained above, with the precise detection of intermediate shocks between release and drop, non-invasive diagnostics can be performed on components that generate noise through shocks or friction.Furthermore, as mentioned above, the method according to the invention can be applied to authentication based on the analysis of the acoustic signal emitted by the watch movement. The method makes it possible to distinguish between the signature of the watch movement and the signature of the case, thus enabling advanced authentication, particularly by differentiating the influences of various components extracted from the signal, and thereby authenticating different components of the watch movement or timepiece. Finally, as mentioned above, the method according to the invention can be applied to diagnostics (excluding chronometry) based on the analysis of the acoustic signal emitted by the watch movement or timepiece.

[0109] Throughout this document, the timepiece referred to may be a watch, specifically a wristwatch.

Claims

Demands:

1. A method for determining a characteristic, in particular a chronometric characteristic, of a watch movement (100) or a timepiece (200) comprising: - a step of acquiring a signal emitted by the watch movement (100) or the timepiece (200), - a step of decomposing the acquired signal in order to obtain a decomposed signal using a transformation into discrete wavelets expressed in a basis, in particular into wavelets of the Daubechies or Symlets type, - a filtering step of the decomposed signal to obtain a filtered signal, - a step of analyzing the filtered signal in order to obtain the characteristic.

2. Method of determination according to claim 1, characterized in that the signal emitted is an acoustic signal emitted by the watch movement (100), in particular by an escapement of the watch movement (100), or emitted by the watch part (200).

3. Method of determination according to claim 2, characterized in that the signal emitted is an acoustic signal emitted by an escapement during its operation and / or a regulating organ during its operation or by an automatic winding system or by a manual winding system or by a date drive system or by a chronograph or by a clasp or by a rotating bezel.

4. Method of determination according to claim 2, characterized in that the signal emitted is an acoustic signal emitted by one or more components of the watch movement or of the timepiece during the operation of the timepiece.

5. A method for determining according to any one of the preceding claims, characterized in that the analysis step includes detecting the shock instants of an escapement, in particular the intermediate shocks between the impulse and the fall, and using these instants to determine a chronometric characteristic.

6. Method of determination according to any one of the preceding claims, characterized in that the analysis step includes a determination of the characteristics and / or times of all shocks of an escapement.

7. Method of determination according to any one of the preceding claims, characterized in that the analysis step includes a determination of the amplitude of oscillation of a pendulum of an oscillator without optical measurement of a lift angle value.

8. Method of determination according to any one of claims 1 to 7, characterized in that the analysis step includes a determination of the step or the reference mark.

9. Method of determination according to one of the preceding claims, characterized in that the filtered signal or parameters of the filtered signal constitute an acoustic signature of a watch movement (100) or of a watch part (200).

10. Method of determination according to any one of the preceding claims, characterized in that the filtered signal is used to determine a signature of components of a watch movement (100) or of a watch part (200).

11. Method of determination according to the preceding claim, characterized in that the filtered signal is used to determine the nature of the materials of components of a watch movement (100) or of a watch part (200).

12. Method of determination according to any one of the preceding claims, characterized in that the filtered signal is used to perform a diagnosis or authentication of a component of a watch movement (100) or of a component of a timepiece (200) or of a watch movement (100) or of a timepiece (200) by comparison with reference and / or previous measurements on the same watch movement (100) or on the same timepiece (200) or on the same model of watch movement (100) or on the same model of timepiece (200).

13. Method of determination according to one of the preceding claims, characterized in that the filtered signal is used to identify natural frequencies of the blanks (190) of a watch movement (100) or of a case (103) of a watch part (200).

14. Method of determination according to any one of claims 1 to 13, characterized in that the filtered signal is used to identify natural frequencies of components of a watch movement (100) or of components (101, 102, 103, 110, 120, 130, 140, 150) of a watch part (200).

15. Diagnostic or authentication method, comprising: - a step in implementing the determination method according to one of the preceding claims applied to a watch movement (100) alone, and / or - a step of casing the watch movement (100) and a step of implementing the determination method according to one of the preceding claims applied to the cased watch movement (100).

16. Diagnostic or authentication method, comprising: - a step in implementing the determination method according to one of claims 1 to 14 applied to a watch movement (100) or to a watch part (200), - a step of applying a controlled mechanical stress to the watch movement (100) or to the watch part (200), in particular by a controlled element for applying the controlled mechanical stress, - a step of acquiring and analyzing the signal emitted by the watch movement (100) or by the watch part (200) in response to the controlled mechanical stress.

17. Diagnostic or authentication method according to claim 15 or 16, characterized in that it comprises a training phase in which the steps of claims 15 or 16 are implemented and in which the results of the analyses are processed by machine learning, for example of the neural network type.

18. Determination device (1) comprising means (2, 21, 22, 23, 3) for implementing the method according to one of the preceding claims.

19. Product computer program downloadable from a communication network and / or recorded on a data medium readable by a computer and / or executable by a computer, characterized in that it includes instructions which, when the program is executed by the computer, lead the computer to implement a process according to any one of claims 1 to 17.

20. Computer-readable recording medium (22) comprising instructions which, when executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 17.

21. Signal from a data carrier, carrying the computer program product according to claim 19.

Images