Method for determining hrtf, and hearing aid

Description

[0001] The invention relates to a method for determining an HRTF in connection with a hearing aid. HRTF stands for "head-related transfer function".

[0002] A hearing aid generally serves to output audio signals to a user. For this purpose, the hearing aid has at least one receiver (also called a loudspeaker or receiver) which converts the audio signal into a sound signal. The hearing aid is typically worn by the user in or on the ear. In one possible configuration, the hearing aid is specifically designed to provide hearing assistance to a user with hearing loss. For this purpose, the hearing aid has a microphone that picks up sound signals from the environment and generates an audio signal, which is an electrical input signal. This signal is then used by the hearing aid's signal processing system to modify the signal. This modification is carried out, for example, based on the user's individual audiogram, which is assigned to the hearing aid, thus compensating for the user's specific hearing deficit.The signal processing results in an electrical output signal, which is a modified audio signal and is then converted back into a sound signal via the hearing aid receiver and output to the user.

[0003] A hearing aid is either monaural, in which case it is worn on only one side of the head, or binaural, in which case it consists of two separate devices worn on opposite sides of the head. Depending on the type, the hearing aid is worn on, in, or behind the ear, or a combination thereof.

[0004] Common types of hearing aids include BTE, RIC, and ITE hearing aids. These differ primarily in their design and how they are worn.

[0005] The HRTF (Head-Related Transmittal Function) is a transfer function that describes how sound signals from the environment are modified by a person's body shape, specifically head shape, as they travel to the ear canal. The HRTF is a transfer function specifically for sound signals, i.e., acoustic signals. In a hearing aid, the HRTF is appropriately used in signal processing to modify these signals, enabling, in particular, the preservation or creation of spatial sound information (so-called "spatial cues"), allowing the user to better localize the sound source.

[0006] In general, a sound signal from an audio source spreads through the environment and thus reaches the ear and ear canal of a hearing aid user. The path along which the sound signal enters the ear canal is also called the acoustic path. The modification of the sound signal along the acoustic path depends on the user's body, specifically the shape of the torso and head, and especially the shape of the ear, particularly the pinna. Consequently, the actual HRTF is usually individual and different for each user. Typically, however, a non-individual HRTF is used, which is determined, for example, with the help of a dummy (e.g., KEMAR) and then applied to a large number of potentially different users.However, this regularly fails to adequately take into account the individual body shape of the user; in any case, possible deviations from the dummy remain unconsidered.

[0007] In principle, it is conceivable to determine an individual HRTF for each user. For this purpose, the user is placed in a room with minimal anechoic conditions and exposed to sound signals from various directions. Several loudspeakers are positioned around the user at fixed, predetermined locations. A microphone is placed where the hearing aid receiver will later sit, i.e., in or on the user's ear, to receive the sound signals. By comparing the transmitted and received sound signals, the individual HRTF can then be determined. This method yields very good results but is also very complex.

[0008] US Patent 9,591,427 B1 describes a method performed by a smartphone to generate HRTFs (Heart Rate Functions) of a person wearing headphones. Using a camera in the smartphone, the position of the smartphone relative to the person's face is determined based on an image of the person's face. While the smartphone is in the person's hand and close to their face, a sound is generated by the smartphone, and its position relative to the person's face is also recorded. The sound is then captured by the left microphone of the headphones in the person's left ear and by the right microphone of the headphones in the person's right ear. Finally, the smartphone generates a left and a right HRTF.

[0009] Reference is also made to US 2015 / 289063 A1, US 2019 / 110137 A1 and US 2016 / 277865 A1.

[0010] Against this background, an object of the invention is to determine the HRTF in a user-specific manner, taking into account the individual body shape of the user. The determination of the HRTF should be as simple as possible and should not disturb the user. To this end, a method for determining the HRTF for use with a hearing aid will be described. Furthermore, a suitable hearing aid will be described.

[0011] The problem is solved according to the invention by a method with the features of claim 1. Advantageous embodiments, further developments, and variants are the subject of the dependent claims. The descriptions relating to the method also apply mutatis mutandis to a hearing aid and vice versa. Where steps of the method are described below, advantageous embodiments for the hearing aid result in particular from the fact that the hearing aid has a control unit configured to perform one or more of these steps.

[0012] A key concept of the present invention is, in particular, to use an audio source capable of both acoustic and non-acoustic output to determine a specific user's HRTF. The audio source is preferably a media device, specifically a device for outputting and / or playing back media (e.g., audio, video). The audio source is used by the user regularly in their daily life. The acoustically output audio signal propagates along an acoustic path to the user, specifically to a microphone in the user's hearing aid, and is modified along this path by the user's body shape. This modification is defined by a transfer function that corresponds to the user's actual, individual HRTF.The non-acoustically output audio signal, however, is not modified by this HRTF, so that by comparing the two differently transmitted audio signals, the HRTF can be individually determined for the user. This is implemented here in order to then determine the user-specific HRTF. At the same time, the method is advantageously applicable with a wide variety of audio sources in the user's everyday life and during the intended use of the hearing aid, thus requiring no special measurement environment or equipment and causing minimal or no disturbance to the user.

[0013] The procedure described here is generally used to determine an HRTF (i.e., "head-related transfer function"). The HRTF is particularly relevant for the intended operation of a user's hearing aid. Advantageously, the HRTF is determined specifically for that user. "Determination" in this context means that the HRTF is ascertained or measured. The HRTF is either determined from scratch or based on a non-user-specific base HRTF, which is then adjusted and preferably optimized within the procedure to obtain a user-specific HRTF. The HRTF is, for example, a parameterized function with one or more parameters that are selected and / or adjusted during the HRTF determination process.

[0014] In this method, an audio source outputs a source audio signal, both acoustically as a sound signal and non-acoustically as a data signal. The source audio signal is an audio signal and, as such, specifically an electrical signal. The source audio signal is also referred to as the "original audio signal." For the acoustic output of the source audio signal, the audio source has a loudspeaker, i.e., an electroacoustic transducer, with which the source audio signal is converted into a sound signal and output. The same source audio signal is also output on another, non-acoustic channel, namely as a data signal. For the non-acoustic output of the source audio signal, the audio source has a data output, which outputs the source audio signal as a data signal and, if necessary, converts it into a suitable data format. Preferably, the data output is an antenna for a wireless communication (e.g.,Bluetooth or WiFi), so that the data signal is transmitted wirelessly. However, a wired connection is also conceivable and suitable; the data output would then be a corresponding connection (e.g., an audio jack or USB port). The essential point is simply that the same audio signal (namely, the source audio signal) is output on two different channels: once acoustically as a sound signal and once non-acoustically (e.g., electrically, electromagnetically, optically) as a data signal.

[0015] The sound signal is received by the hearing aid and converted back into an audio signal, namely a first audio signal. This first audio signal is also referred to as an "acoustically transmitted audio signal" because it results from the source audio signal through conversion into a sound signal and back again. Specifically for a hearing aid designed for a hearing-impaired user, receiving sound signals from the environment is a core function of the hearing aid.

[0016] The data signal is received by the hearing aid or by another device, which generates a second audio signal from the data signal, particularly via a data input, such as an antenna. The other device—if present—is typically an accessory connected to the hearing aid for data exchange, for example, via Bluetooth or Wi-Fi. This other device could be a smartphone, for instance. While it is theoretically possible for the audio source itself to be the other device, for the purposes of this discussion, it is assumed without limitation that this is not the case. The second audio signal is also referred to as a "non-acoustically transmitted audio signal" because it is derived from, or even identical to, the source audio signal without having been acoustically transmitted.Apart from the fact that the sound signal and the data signal are physically transmitted in two different ways, they are preferably also transmitted in different frequency ranges. The sound signal is transmitted primarily in the audible frequency range of 20 Hz to 20 kHz, while the data signal is transmitted in a communication frequency range, e.g., between 1 MHz and 10 GHz, and in any case at a frequency several orders of magnitude higher.

[0017] The first and second audio signals are each a transmitted (and therefore possibly modified) version of the source audio signal. For the sake of completeness, the source audio signal is also referred to as the "third audio signal".

[0018] The first and second audio signals, i.e., the audio signals transmitted on different channels, are compared, and the HRTF is determined based on this comparison. The underlying assumption is that the second audio signal typically closely matches the source audio signal and has not been affected by the HRTF. In contrast, the sound signal has been modified by the HRTF, so the first audio signal differs from the source audio signal accordingly. Therefore, as a first approximation, the following relationship exists between the first, acoustically transmitted audio signal A1 and the second, non-acoustically transmitted audio signal A2: A2 = HRTF(A1). The precise method of comparison is essentially irrelevant.More importantly, the data signal provides an audio signal unaffected by the HRTF, which is used as a reference signal to determine the actual, user-specific HRTF.

[0019] In a suitable implementation, the first audio signal is used as a target signal and the second audio signal as an actual signal to determine the HRTF. In this way, the HRTF is determined based on the difference between the audio signal and the data signal (more precisely, based on the difference between the first and second audio signals). How the HRTF is specifically calculated is initially secondary and depends primarily on how the HRTF is parameterized, i.e., which parameters define it. In principle, it is possible to perform numerical optimization with sufficient computing power. Here, individual parameters (also called coefficients) are varied until a minimal deviation is reached (i.e., a minimum or at least a stable and possibly only local minimum). A suitable optimization algorithm is, for example, LASSO (i.e., "least absolute shrinkage and selection operator").

[0020] The HRTF determined in the aforementioned manner is stored in the hearing aid and preferably used by the hearing aid's signal processing during operation to adjust the sound signal output to the user. The specific use of the HRTF is not important here. Possible uses include generating acoustic cues with spatial information to aid navigation for a walking user, adding spatial information to a streaming signal so that it sounds to the user as if it is coming from an audio source, such as a television, in a specific direction, and virtual controls where, for example, the position of a control element, such as a slider, is represented acoustically with a spatial effect.Emphasis on the right or left side depending on the position of the control element). Especially in connection with "in-ear" headphones, the use of an HRTF to modify the audio output is advantageous.

[0021] In a specific, suitable implementation, only excerpts, so-called samples, are taken from the first and second audio signals to determine the HRTF and stored as a single data set. The two excerpts (one from the first audio signal and one from the second) of each data set preferably originate from the same time interval or have a matching timestamp. This ensures that the HRTF is correctly determined by comparing the two sections. Typically, a large number of such data sets are recorded and stored and evaluated to determine the HRTF. This is done either on the hearing aid, on an additional device as described, or on a separate computer, e.g., a server.

[0022] As already mentioned, an HRTF is preferably parameterized, meaning it's a function with a number of parameters that can vary depending on the user. When determining the HRTF, these parameters are specifically optimized and thus adapted to the user. The HRTF is preferably determined continuously, so that the HRTF used increasingly approximates the actual, individual HRTF over time. The process is therefore iterative. This also advantageously takes changes in the user's body shape into account.

[0023] Without limiting the generality of the invention, it is assumed here that the hearing aid is designed for use by a hearing-impaired user. However, the invention is also applicable to other hearing aids, such as headphones, which additionally have one or more microphones. A hearing aid for a hearing-impaired user generally comprises an input transducer, a signal processor, and an output transducer. The input transducer is a microphone and serves to pick up sound signals from the environment, i.e., also to receive the sound signal emitted by the audio source. The output transducer is usually a receiver, also referred to as a loudspeaker or receiver. While it is assumed here, without limiting the generality of the invention, that the hearing aid has a receiver, other output transducers for output to the user are also suitable.The hearing aid is typically assigned to a single user and used exclusively by that user. The input converter generally generates an input signal, which is then fed into the signal processing unit. In this case, the input converter specifically generates the initial audio signal, which is therefore an input signal. The signal processing unit modifies the input signal, thereby generating an output signal, which is thus a modified input signal. To compensate for hearing loss, the input signal is amplified, for example, according to the user's audiogram, using a frequency-dependent gain factor. Alternatively or additionally, the input signal is modified according to the HRTF (Hearing-Related Frequency Factor). Finally, the output signal is delivered to the user via the output converter.

[0024] The previously described reception and playback of a sound signal with electrical modification is the standard operating mode of the hearing aid; this is also referred to as "normal operation." In addition to normal operation, the hearing aid described here preferably also features a streaming mode, in which the output to the user is based on the data signal transmitted by the audio source. Streaming mode has the advantage that conversion to and from a sound signal is unnecessary and preferably avoided, allowing the audio signal to be transmitted from the audio source to the user without loss or alteration. Streaming mode is used, for example, to transmit an audio signal from a television, computer, or smartphone—or generally from any audio source—to the hearing aid.The hearing aid accordingly has a data input, which is designed to complement the data output of the audio source, preferably also as an antenna. The specifications for the data output apply analogously to the data input and vice versa. The hearing aid is suitably designed so that the user can switch between normal operation and streaming operation.

[0025] With headphones or similar devices, the normal operation described above may not apply, and streaming operation is the norm.

[0026] The functionalities of normal operation and streaming operation are advantageously combined here to determine the HRTF. On the one hand, the hearing aid receives the sound signal from the audio source via the microphone, thus utilizing the functionality of normal operation. On the other hand, the hearing aid receives the data signal from the audio source, thus utilizing the functionality of streaming operation. Which of the two audio signals (first and second audio signal) is actually output to the user via the receiver is irrelevant and is conveniently left to the user's discretion. For the procedure described here, the only relevant factor is that both audio signals are present in order to determine the HRTF based on them.

[0027] Furthermore, for the procedure described here, it is not strictly necessary for the hearing aid to have streaming capabilities or even to receive the data signal at all; it can also be received by another device. The first and second audio signals simply need to be combined on some device to be compared and the HRTF determined based on this comparison. While the hearing aid is generally suitable for this purpose, a computer, especially a server, is equally suitable, as it typically offers significantly higher processing power than a hearing aid. It is also conceivable that the hearing aid receives the data signal, but that the HRTF determination is not performed by the hearing aid itself, but rather by, for example, a smartphone or server to which the hearing aid transmits the audio signals or the corresponding data sets.

[0028] However, for the correct determination of the HRTF, it is crucial that the hearing aid receives the sound signal. This is because the hearing aid is worn by the user, whereas any other device is typically positioned away from the user and therefore unsuitable for receiving a sound signal propagating along the acoustic path to the user's ear. Accordingly, in a preferred embodiment, the hearing aid receives the sound signal with a microphone that is an integral part of the hearing aid. Optionally, the hearing aid may even have multiple microphones with which the sound signal is received and the initial audio signal is generated. The hearing aid is advantageously designed such that, when worn, the microphone is positioned in or on one of the user's ears. Specifically, the microphone is thus positioned behind the ear, in the ear, or in the ear canal. The exact position of the microphone depends on the type of hearing aid.With a BTE device, the microphone is positioned behind the ear; with a RIC device, it is in the ear canal; and with an ITE device, it is in the ear but in front of the ear canal. Therefore, the entire acoustic path into the ear canal may not be considered, and the HRTF may only be determined for a portion of the acoustic path—that is, for one or more, but not all, segments.

[0029] The hearing aid is either monaural, in which case it is worn on only one side of the head (left or right), or binaural, in which case it consists of two separate devices worn on opposite sides of the head (i.e., left and right). In a binaural hearing aid, each of the two devices has one or more microphones for receiving sound signals.

[0030] Preferably, the determination of the HRTF also includes a spatial situation with respect to the user. This spatial situation is preferably selected from a set of spatial situations, comprising and in particular consisting only of: the user's position relative to the audio source, the user's distance relative to the audio source, the user's orientation relative to the audio source, the orientation of the user's head relative to their torso, and the user's posture. The orientation of the user's head relative to their torso is a specific posture; other postures include, for example, sitting, lying down, and standing. The orientation of the head relative to the torso is preferably a head rotation about the user's longitudinal body axis, a head tilt about the user's transverse axis (i.e., a forward / backward nod), or lateral flexion (i.e., a head tilt to one side).

[0031] In a suitable configuration, a corresponding spatial situation is determined and taken into account when calculating the HRTF. This is based on the consideration that the acoustic path regularly depends on how the user's body is oriented relative to the audio source and / or what posture the user assumes, i.e., whether the sound signal reaches the user from the front, back, or side, and how the user's own body, especially the torso, attenuates the sound signal. Accordingly, the modification of the sound signal during its propagation to the user's ear depends on the relative spatial relationship between the user and the audio source as well as the user's posture, so that the HRTF is generally situation-dependent and specifically direction-dependent and posture-dependent.To determine the HRTF as optimally as possible, it is advantageous not only to record as many data sets as possible in general, but also to record data sets for as many spatial situations as possible, i.e., in as many different relative spatial relationships of the user to the audio source and / or for as many body postures of the user as possible. This allows the HRTF to be determined in a situation-dependent and specifically direction-dependent and / or posture-dependent manner.

[0032] How exactly the spatial situation is determined is of secondary importance here and therefore not the subject of further discussion; in principle, any known method is suitable. In a suitable configuration, the hearing aid is a binaural hearing aid and accordingly receives the sound signal from the audio source on both sides. The user's orientation relative to the audio source is then determined, for example, based on a time delay or amplitude difference of the sound signal received on the two sides. Tracking the user, for example, using a camera of the audio source or a beacon in an accessory worn by the user, is also conceivable and suitable. Another suitable configuration involves determining the absolute location of both the audio source and the hearing aid, and then calculating the relative spatial relationship by differentiating these locations.The orientation of the head is determined, for example, by means of video observation of the user, by means of a gyroscope or magnetometer, especially of the hearing aid, or it is assumed that the orientation is "looking straight ahead" if there has been no change in orientation for a longer period of time (e.g. at least 1 minute).

[0033] Suitablely, to determine the HRTF, a respective excerpt from the first and second audio signals and a spatial situation relating to the user are stored together as a single data set. Each data set then contains not only a sample of each of the two audio signals, but also additional information about the user's relative spatial relationship to the audio source and / or the user's posture at the time of these samples.

[0034] Data sets can be generated in a variety of ways, especially with varying degrees of user involvement and with or without specific control of the audio source.

[0035] A suitable design is one in which data sets are continuously generated without requiring any user interaction or specific control of the audio source. In this way, the process runs in the background during normal use and therefore does not disturb the user.

[0036] In a suitable configuration, the audio source is controlled such that, when a spatial situation exists with respect to the user for which a minimum number of data records do not yet exist, it outputs a source audio signal in order to generate a data record for that spatial situation. In this configuration, the audio source is thus specifically controlled to generate a data record for those spatial situations for which there are not yet enough data records for a sufficiently accurate determination of the HRTF. The actual number of data records required for a given spatial situation, i.e., the minimum number, is initially irrelevant. For example, the minimum number could be just 1, or alternatively 10, 100, or 1000.Even in this configuration, user participation is not required; however, the audio source is specifically controlled to generate the most relevant data sets. The hearing aid or other device checks, for example, the current position, distance, orientation, and / or posture, and whether the number of existing data sets meets at least the minimum requirement. If not, the audio source is controlled accordingly to output the source audio signal as both a sound signal and a data signal, thus generating a data set for the current position, distance, orientation, and / or posture.

[0037] In a suitable implementation, the user is given an instruction to create one or more spatial situations. The audio source then outputs a source audio signal for each of these situations to generate a data set. The instruction is issued, for example, by the hearing aid, the audio source, or another device. The instruction can be, for instance, acoustic or visual. Whether the user actually follows the instruction is entirely up to them. However, there is a probability that the user will create the required spatial situation in response to the instruction, so that a data set can then be generated for it, and indeed will be. The process thus utilizes user participation; however, specific control of the audio source is not strictly necessary.

[0038] The hearing aid in question features a test mode in which it outputs a signal to the user containing spatial sound information (i.e., a "spatial cue," e.g., a spatially localized sound) to prompt the user to move or orient themselves in a predetermined direction (either entirely or just with their head), specifically towards the perceived source of the sound. The actual direction in which the user moves or orients themselves is then determined and compared to the intended direction to assess the degree of adaptation. HRTF to determine the user. The degree of adaptation indicates, in particular, how well the currently determined HRTF with the actual HRTF matches. This is verified in test mode by using the currently determined HRTF the output signal is generated. If the currently determined HRTF from the actual HRTF If the sound is misinterpreted, the user will incorrectly locate it in a different direction than if it were actually coming from the intended direction. HRTF would be modified. The test mode thus allows for a review of the previously determined HRTF and also a determination of how well it matches the actual HRTF for the user. In one example configuration, the user is instructed to look in a specific direction with neutral pupils. The data sets obtained in this way are then used to test the degree of adaptation. This approach is less demanding for the user than having them move around the room. Additionally, this method is useful for collecting data on missing spatial situations regarding the orientation of the head relative to the torso.

[0039] The HRTF is fundamentally decomposable into several individual transfer functions, each modeling a specific section of the acoustic path. These individual sections, when combined, yield the HRTF for the entire acoustic path. However, it may not be necessary, or even possible, to determine the HRTF for the entire acoustic path in this manner, but rather only for one or more individual sections, specifically those closest to the user. The remaining sections are then modeled using their respective standard functions.

[0040] In a practical implementation, the HRTF is determined starting from a base HRTF, which is a transfer function for only the first segment of an acoustic path from the audio source to the user's ear canal. The HRTF is then primarily determined for a different, second segment of the acoustic path. This approach is based on the consideration that the HRTF is typically most strongly defined by the user's ear, and specifically their pinna, and less so by the user's torso or general head shape. Therefore, the second segment primarily includes that part of the acoustic path containing the pinna. The base HRTF is then, for example, the HRTF of a dummy and primarily considers the user's body shape and general head shape.This basic HRTF is then optimized by the present method to take into account the specific shape of the user's pinna, thus determining the HRTF in a user-specific manner. For this purpose, the hearing aid is advantageously designed so that its microphone, when worn, is positioned in the ear canal or in the ear of the user, and not merely behind the ear.

[0041] As already mentioned, the HRTF is not necessarily determined by the hearing aid. Preferably, the HRTF is determined by a computer, in particular a server, which is separate from the hearing aid and the audio source. For the purposes of this discussion, we will assume a server as the computer, without limiting the generality of the explanation. The precise method by which the data reaches the server is not particularly relevant and depends on the chosen methodology, the hearing aid, the audio source, and any other devices involved. For example, the hearing aid sends the first acoustically transmitted audio signal, or excerpts thereof, to the server. Similarly, the second acoustically transmitted audio signal, or excerpts thereof, is sent to the server by the hearing aid or another device, such as a smartphone or the audio source. The server then conveniently sends the HRTF to the hearing aid.

[0042] Preferably, the audio source is a stationary device. "Stationary" here refers specifically to being unmoved, but not necessarily immobile in general. In other words, the audio source typically remains in the same location within an environment, such as a room, while the user moves relative to the audio source, and the spatial situation in relation to the user generally changes. A stationary device has the particular advantage that any movement by the user automatically creates a change in the spatial situation, thus making it easy to generate data sets for different spatial situations.

[0043] A particularly preferred configuration uses a television set as the audio source. A television set is typically a stationary device. Using a television set as the audio source in a method like the one described here offers several advantages. Firstly, a television set typically has one or more speakers with high output quality, covering a particularly wide frequency spectrum and reproducing the source audio signal with exceptional fidelity. This is especially true compared to a smartphone. Furthermore, the user is typically positioned a few meters away from the television set, a distance similar to that used when determining an HRTF in an anechoic chamber, as described earlier, and which is optimal for determining the HRTF.Furthermore, the TV is typically always placed in the same position within the environment, allowing additional room acoustic effects to be better considered when determining the HRTF, particularly along a section of the acoustic path not represented by the HRTF. Finally, it is also expected that sound signals from a TV will not contain sensitive, personal data, unlike, for example, sound signals from a smartphone.

[0044] Preferably, the procedure described here is carried out while the user is watching television, particularly with the audio source, i.e., while the audio source, which is a television set, is switched on and the user is in its immediate vicinity (e.g., within 5 meters of the audio source). It is not essential that the user follows or pays special attention to the content broadcast by the television set. Carrying out the procedure while the user is watching television has several advantages.

[0045] On the one hand, it is expected that the user will watch television for an extended period, e.g., 1 to 2 hours, resulting in a correspondingly large number of data records. It is also expected that the user will watch television repeatedly, thus ensuring a consistent volume of data records. Furthermore, the user typically does not engage in personal conversations with other people while watching television, ensuring that no sensitive personal data is recorded. If it is, it is appropriately discarded. A personal conversation is recognized by the hearing aid, for example, by the fact that the corresponding sound signal arrives from a different direction than the sound signal from the audio source. Other background noise is also typically absent during television viewing, as the user regularly switches off other noise sources, resulting in overall data records of very high quality.

[0046] Using a television is also advantageous because, although it typically has multiple speakers, it can also be operated in such a way that only a single speaker is used to output sound signals. This makes the determination of the HRTF significantly more accurate, since there is now only a single sound source and the acoustic path is thus very precisely defined. This also applies generally to all audio sources with multiple speakers. In an advantageous embodiment, the audio source is therefore controlled in such a way that it outputs the audio signal as a sound signal via only a single speaker. Output via only one speaker is not limiting for the user, at least insofar as they can advantageously also receive the source audio signal as a data signal via streaming and are therefore not dependent on the sound output of the audio source.Preferably, the hearing aid is operated in streaming mode during the procedure. If the additional sound output is perceived as disturbing, the sound signal is filtered out, for example, by the hearing aid using an ANC unit (ANC stands for "active noise cancelling").

[0047] In a suitable configuration, an acoustic parameter of the environment is determined, particularly to quantify one or more room acoustic effects, and taken into account when determining the HRTF. Room acoustic effects include, for example, reflections of the sound signal from walls or objects in the environment, or reverberation, especially in a room. Accordingly, acoustic parameters of the environment are a time or amplitude that quantifies an impulse response of the environment, an early reflection, or reverberation. The acoustic parameter is determined, for example, with the hearing aid or another device. It is also useful to place an additional microphone in the room to determine the acoustic parameter.

[0048] A hearing aid has a control unit which is designed to perform a procedure as described above, possibly in combination with an audio source and / or another device as described.

[0049] The task will continue to be solved primarily by a computer and / or another device, e.g. a smartphone, as described above.

[0050] Exemplary embodiments of the invention are explained in more detail below with reference to a drawing. Each drawing schematically shows: Fig. 1 an environment with an audio source and a user with a hearing aid, Fig. 2 an acoustic path, Fig. 3 a hearing aid, Fig. 4 the determination of an HRTF from multiple data sets, Fig. 5 an audio source, a hearing aid and a computer.

[0051] In Fig. 1 This illustrates a key concept of the present invention, namely, to determine an HRTF 2 for a specific user 4 by using an audio source 6 that can output a source audio signal 8 both acoustically and non-acoustically. The audio source 6 is a media device, specifically a television. The audio source 6 is used repeatedly by the user 4 in their daily life. The acoustically output audio signal propagates along an acoustic path 10 to the user, specifically to a microphone 12 of a hearing aid 14 belonging to the user 4, and is modified along the acoustic path 10 by the user 4's body shape.

[0052] An exemplary acoustic path 10 is in Fig. 2 The diagram shows several sections 16, 18, and 20. A first section 16 is defined by an initial modification, which occurs independently of user 4 due to the environment, and is not relevant here. A second section 18 is defined by a second modification, which occurs due to the body shape (primarily torso shape) and head shape of user 4. The second section 28 forms the acoustic path 10 over / along / through the body of user 4 to the ear or behind the ear of user 4. A third modification 20 occurs through the ear, specifically the pinna of user 4, and thus defines a third and final section 20 of the acoustic path 10 from outside the ear to the ear canal of user 4. Sections 18 and 20 are defined by a transfer function that corresponds to the actual, individual HRTF of user 4.The non-acoustically output audio signal, on the other hand, is not modified by this HRTF 2, so that by comparing the two differently transmitted audio signals, the HRTF 2 can be individually determined for the user.

[0053] The method described here is generally used to determine an HRTF 2 (i.e., "head-related transfer function"). The determination of the HRTF 2 is user-specific for a particular user 4. The audio source 6 outputs the source audio signal 8, both acoustically as a sound signal 22 and non-acoustically as a data signal 24. The source audio signal 8 is an audio signal and, as such, an electrical signal. For acoustic output of the source audio signal 8, the audio source 6 has a loudspeaker 26. The same source audio signal 8 is also output on another, non-acoustic channel, namely as a data signal 24. For non-acoustic output of the source audio signal 8, the audio source 6 has a data output 28, in the illustrated embodiment an antenna for a wireless connection. However, wired transmission is also possible; the data output 28 then serves as a corresponding connection.The essential point is simply that the same source audio signal 8 is output on two different channels, namely once acoustically as a sound signal 22 and once non-acoustically as a data signal 24.

[0054] The sound signal 22 is received by the hearing aid 14 and converted back into an audio signal, namely a first audio signal 30, which is also referred to as an "acoustically transmitted audio signal". Specifically, in the case of a hearing aid 14 for a hearing-impaired user 4, receiving sound signals 22 from the environment is a core function of the hearing aid 16. The data signal 24 is received by the hearing aid 14 or by another device 32, which generates a second audio signal 34 from the data signal 24. For this purpose, the hearing aid 14 or the other device 32 has a corresponding data input 44, e.g., an antenna. In the illustrated embodiment, the other device 32 is an accessory device connected to the hearing aid 14 for data exchange, for example, a smartphone. The second audio signal 34 is also referred to as a "non-acoustically transmitted audio signal".

[0055] The first audio signal 30 and the second audio signal 34, i.e., the audio signals transmitted on different channels, are compared, and the HRTF 2 is determined based on this comparison. The second audio signal 34 typically largely corresponds to the source audio signal 8 and has not been affected by the HRTF 2. In contrast, the sound signal 22 has been modified by the HRTF 2, so the first audio signal 30 differs accordingly from the source audio signal 8. To determine the HRTF 2, the first audio signal 30 is then used, for example, as a target signal, and the second audio signal 34 as an actual signal.

[0056] The HRTF 2 determined in the aforementioned manner is stored in the hearing aid 14 and used by a signal processing unit 36 ​​of the hearing aid 16 during operation to adjust the sound signal that is output by the hearing aid 14 to the user 4. An example hearing aid 14 is shown in Fig. 3 The hearing aid 14 shown here is, without limitation of generality, a hearing aid 14 for the provision of hearing to a hearing-impaired user 4. The invention is also applicable to other hearing aids 16, such as headphones, which additionally have one or more microphones. The hearing aid 14 shown here has an input transducer (namely the microphone 12), the aforementioned signal processing unit 36, and an output transducer 38, here a receiver. The input transducer generates an input signal, which is fed to the signal processing unit 36. In this case, the input transducer specifically also generates the first audio signal 30, which is accordingly an input signal. The signal processing unit 36 ​​modifies the input signal and thereby generates an output signal, which is thus a modified input signal.To compensate for hearing loss, the input signal is amplified, for example, according to the audiogram of user 4, using a frequency-dependent gain factor. Alternatively or additionally, the input signal is modified depending on the HRTF 2. The output signal is then sent to user 4 via the output converter 38.

[0057] In the embodiment shown here, to determine the HRTF 2, only excerpts 40, so-called samples, are taken from the first and second audio signals 30 and 34 and stored as a data set 42. This is in Fig. 4 Illustrated. The two excerpts 40 (one excerpt 40 from the first audio signal 30 and one excerpt 40 from the second audio signal 34) of a respective data set 42 also originate from the same time interval or have a matching timestamp. Typically, a large number of data sets 42 are recorded and stored and evaluated to determine the HRTF 2. This is done either on the hearing aid 14, on an additional device as described, or on a separate computer, e.g., a server.

[0058] The previously described reception and re-emission of a sound signal with electrical modification is the standard operating mode of hearing aid 16; this is also referred to as "normal operation" of hearing aid 16. In addition to normal operation, the hearing aid 14 described here also features a streaming mode, in which the output to user 4 is based on the data signal 24 emitted by audio source 6. During streaming operation, no conversion to or from a sound signal is performed, and an audio signal is transmitted from audio source 6 to user 4 without loss or alteration. Streaming operation is used, for example, to transmit an audio signal 8 from a television, computer, or smartphone, and generally from any audio source 6, to hearing aid 14.The hearing aid 14 accordingly has a data input 44, which is designed to be complementary to the data output 28 of the audio source, thus also acting as an antenna.

[0059] The functionalities of normal operation and streaming operation are now combined to determine HRTF 2. Hearing aid 14 receives the sound signal 22 from audio source 6 via microphone 12, thus utilizing the functionality of normal operation. Simultaneously, hearing aid 14 receives the data signal 24 from audio source 6, thus utilizing the functionality of streaming operation. Which of the two audio signals 30, 34 is actually output to user 4 via output converter 38 is irrelevant and can be determined, for example, by the user.

[0060] However, for the procedure described here, it is not absolutely necessary that the hearing aid 14 has a streaming function or generally receives the data signal 24; this can also be received by another device 32. The first and second audio signals 30, 34 simply need to be combined on some device in order to be compared and, based on this, to determine the HRTF 2.

[0061] However, for the correct determination of HRTF 2, it is important that the hearing aid 14 receives the sound signal 22, because the hearing aid 14 is worn by the user 4, while any other device 32 is regularly positioned away from the user 4 and is therefore not suitable for receiving a sound signal 22 propagating along the acoustic path 10 to the user 4. In the embodiment shown here, the hearing aid 14 accordingly receives the sound signal 22 with a microphone 12, which is part of the hearing aid 16. The hearing aid 14 shown here is also designed such that, when worn, the microphone 12 is positioned in or on one of the user 4's ears. The exact position of the microphone 12 depends on the type of hearing aid 16. In a BTE device, the microphone 12 is positioned behind the ear, in a RIC device in the ear canal, and in an ITE device in the ear, but in front of the ear canal.Therefore, if necessary, not the entire acoustic path 10 extending into the ear canal is taken into account, and the HRTF 2 is determined accordingly only for one or individual sections 18, 20 of the acoustic path 10. The hearing aid 14 is either monaural and is then worn only on one side (left or right) of the head, or—as shown here—binaural and then comprises two individual devices, which are worn on opposite sides of the head (i.e., left and right). In a binaural hearing aid 14, both individual devices each have one or more microphones 12.

[0062] In the embodiment shown here, the spatial situation with respect to user 4 is also taken into account when determining HRTF 2, specifically their relative spatial relationship to the audio source 6. In this embodiment, the spatial situation is characterized by the user 4's position 46, distance 48, and / or orientation 50 relative to the audio source 6. In a variant not explicitly shown, the spatial situation is alternatively or additionally specifically the orientation of user 4's head relative to their torso, or more generally, the user 4's posture. Other postures include, for example, sitting, lying down, and standing. The acoustic path 10 regularly depends on how user 4's body is oriented relative to the audio source 6 or on the posture user 4 assumes, i.e., whether the sound signal 22 reaches user 4, for example,from the front, from the back, or from the side, and how the user's own body, specifically their torso, shades the sound signal. Accordingly, the modification of the sound signal 22 during its propagation to user 4 depends on the relative spatial relationship between user 4 and audio source 6, as well as the posture of user 4. Therefore, the HRTF 2 is also situation-dependent, specifically direction-dependent and posture-dependent. For this reason, data sets 42 are recorded in as many different relative spatial situations as possible, i.e., for as many different positions 46, distances 48, orientations 50, and / or postures as possible. How exactly the spatial situation, e.g., the position 46, distance 48, and / or orientation 50 of user 4 relative to audio source 6, is determined is of secondary importance here and is therefore not discussed further.In any case, to determine the HRTF 2, a respective section 40 from the first and the second audio signal 30, 34 and a spatial situation are stored together as a data set 42, so that a respective data set 42 then also contains information about the spatial situation.

[0063] The generation of data sets 42 is possible in a variety of ways, in particular with different degrees of user participation 4 and with or without special control of the audio source 6.

[0064] One possible configuration involves continuously generating data records 42 without requiring any active input from user 4 or specific control of the audio source 6. Thus, when used as intended, the process runs in the background and does not disturb user 4.

[0065] Alternatively or additionally, the audio source 6 is controlled such that, when a spatial situation exists for which a minimum number of data records 42 do not yet exist, it outputs a source audio signal 8 in order to generate a data record 42 for that spatial situation. In this configuration, the audio source 6 is therefore specifically controlled to generate a data record 42 for those spatial situations for which there are not yet enough data records 42 to adequately determine the HRTF 2. The actual number of data records 42 required for a given spatial situation, i.e., the minimum number, is initially irrelevant. For example, the minimum number could be only 1, or alternatively 10, 100, or 1000.Even in this configuration, the participation of user 4 is not required; however, the audio source 6 is specifically controlled in order to generate the most meaningful data sets 42.

[0066] Alternatively or additionally, an instruction is issued to user 4 to create one or more spatial situations in which audio source 6 then outputs a source audio signal 8 to generate a data set 42 for each of these spatial situations. The instruction is issued, for example, by the hearing aid 14, audio source 6, or another device 32. The instruction may be, for example, acoustic or visual. Whether user 4 actually follows the instruction is entirely up to them. The procedure thus utilizes the participation of user 4 overall; however, specific control of audio source 6 is not strictly necessary.

[0067] Alternatively or additionally, the hearing aid 14 has a test mode in which it outputs a signal to user 4 containing spatial sound information (i.e., a "spatial cue," e.g., a spatially localized sound). This signal prompts user 4 to move or orient themselves in a predetermined direction, specifically towards the perceived source of the sound. The actual direction in which user 4 moves or orients themselves is then determined and compared to the intended direction to ascertain the HRTF 2 adaptation level for user 4. The adaptation level indicates, for example, how well the currently determined HRTF 2 matches the actual HRTF. The test mode thus allows for a review of the previously determined HRTF 2 and also a determination of how well it matches the actual HRTF for user 4.

[0068] As in Fig. 2 As is already apparent, the HRTF 2 is fundamentally decomposable into several individual transfer functions, which model individual sections (e.g., sections 18, 20) of the acoustic path 10 and which, when combined, yield the HRTF 2 for the entire acoustic path 10. Under certain circumstances, it is not necessary or even impossible to determine the HRTF 2 for the entire acoustic path 10 in the manner described, but only for one or more individual sections 18, 20, specifically those sections 18, 20 that are closest to user 4, especially the third section 20. The remaining sections 18 are then modeled, for example, using a respective standard function, particularly section 16, which does not itself contribute to the HRTF but whose determination may be distorted.

[0069] In one possible embodiment, the HRTF 2 is determined starting from a basic HRTF, which is a transfer function for only a first section 18, 20 of an acoustic path 10 from the audio source 6 to the user's ear canal 4, so that the HRTF 2 is predominantly determined for another, second section 18, 20 of the acoustic path 10. For example, the second section contains, in particular, that part of the acoustic path 10 which contains the pinna, here the third section 20 in Fig. 2 The basic HRTF is, for example, the HRTF 2 of a dummy and primarily takes into account the body shape and general head shape of user 4. This basic HRTF is then optimized by the present method to consider the specific shape of user 4's pinna, so that the overall HRTF 2 is determined in a user-specific manner. For this purpose, the hearing aid 14 is designed, for example, such that its microphone 12, when worn, is positioned in the ear canal or in the ear of user 4 and not merely behind the ear.

[0070] HRTF 2 is not necessarily determined by hearing aid 14. Fig. 5 The HRTF 2 is determined, for example, by a computer 52, here a server, which is configured separately from the hearing aid 14 and the audio source 6. How exactly the data records 42 reach the server for this purpose is not relevant and also depends on the chosen configuration of the procedure, the hearing aid 14, the audio source 6, and any other devices 32 that may be involved. In this respect, it shows Fig. 5 This is merely one of many possible embodiments. For example, it is possible for the hearing aid 14 to send the first acoustically transmitted audio signal 30 or excerpts 40 thereof to the server, and for the second acoustically transmitted audio signal 34 or excerpts 4 thereof to the server, which then sends the HRTF 2 to the hearing aid 14.

[0071] In the illustrated embodiment of the Fig. 1 The audio source 6 is a stationary device and typically remains in the same location in the environment, e.g., a room as shown, while the user 4 moves relative to the audio source 6, and the spatial situation generally changes. Such movement of the user 4 is in Fig. 1 This is illustrated by an exemplary movement path 54. Furthermore, the audio source 6 in the configuration shown here is a television set. As in Fig. 1As can be seen, user 4 is typically located a few meters away from the TV, a distance similar to that used when determining HRTF 2 in an anechoic chamber as described earlier. Furthermore, the TV is typically always positioned in the same location, allowing for better consideration of additional room acoustic effects, particularly along the first section 16, when determining HRTF 2. The procedure described here is specifically performed while user 4 is watching television, i.e., while audio source 6 is switched on and user 4 is in its immediate vicinity (e.g., within 5 meters of audio source 6). It is not essential that user 4 is actively following or paying particular attention to the content emitted by the TV.In one possible embodiment, the audio source 6 is also controlled in such a way that it outputs the audio signal 8 as a sound signal 22 via only a single loudspeaker 26, so that the acoustic path 10 is defined more precisely.

[0072] Furthermore, in one embodiment, an acoustic parameter of the environment is determined to quantify one or more room acoustic effects and is taken into account when determining the HRTF 2. This results in the determination of a transfer function for the first section 16. Room acoustic effects include, for example, reflections of the sound signal from walls or objects in the environment, or reverberation, especially in a room. The acoustic parameter is determined, for example, with the hearing aid 14 or with another device 32.

[0073] The hearing aid 14 further comprises a control unit 56 which is designed to carry out the procedure as described above, at least those steps of the procedure which are carried out by the hearing aid 14. Reference symbol list

[0074] 2HRTF 4 User 6 Audio Source 8 Source Audio Signal 10 Acoustic Path 12 Microphone 14 Hearing Aid 16 First Section 18 Second Section 20 Third Section 22 Sound Signal 24 Data Signal 26 Speaker (of the audio source) 28 Data Output 30 First Audio Signal (from sound signal) 32 Other Device 34 Second Audio Signal (from data signal) 36 Signal Processing 38 Output Converter 40 Sample 42 Data Set 44 Data Input 46 Position 48 Distance 50 Orientation 52 Computer (Server) 54 Motion Path 56 Control Unit

Claims

1. A method for determining an HRTF (2), a. wherein an audio source (6) outputs a source audio signal (8), namely both acoustically as a sound signal (22) and non-acoustically as a data signal (24), b. wherein the sound signal (22) is received by a hearing aid (14) of a user (4) and is converted by this hearing aid (14) back into an audio signal (30), namely into a first audio signal (30), c. wherein the data signal (24) is received by the hearing aid (14) or by another device (6, 32), which generates a second audio signal (34) from the data signal (24), d. wherein the first audio signal (30) and the second audio signal (34) are compared with one another and the HRTF (2) is determined based thereon, characterized in that e. the hearing aid (14) has a test mode and in this mode outputs an output signal to the user (4), which is generated by means of the currently determined HRTF and which has spatial noise information to prompt the user (4) to move or orient themselves in an intended direction, f. it is determined in which actual direction the user (4) moves or orients themselves and this is compared with the intended direction in order to determine a degree of adaptation of the HRTF (2) to the user (4), wherein the degree of adaptation indicates how well the currently determined HRTF (2) corresponds with an actual HRTF for the user.

2. The method as claimed in claim 1, wherein the hearing aid (14) receives the sound signal (22) using a microphone (12), wherein the hearing aid (14) is designed such that in the worn state, the microphone (12) is positioned in or on an ear of the user (4).

3. The method as claimed in either one of claims 1 to 2, wherein the HRTF (2) is determined by a computer (52), which is formed separately from the hearing aid (14) and the audio source (6).

4. The method as claimed in any one of claims 1 to 3, wherein the audio source (6) is a stationary device.

5. The method as claimed in any one of claims 1 to 4, wherein the audio source (6) is a TV set.

6. The method as claimed in any one of claims 1 to 5, wherein this method is carried out while the user (4) watches television.

7. The method as claimed in any one of claims 1 to 6, wherein the audio source (6) is controlled such that it outputs the source audio signal (8) as a sound signal (22) via only one single loudspeaker (26).

Images