Hearing device and corresponding method for ownvoices detection

p0001The present invention relates to a hearing apparatus, in particular a hearing aid with a microphone for recording an ambient sound from the environment of the user. Moreover, the present invention relates to a corresponding method for operating a hearing device.

p0002The own voice of the hearing aid can not be distinguished from an external sound source in conventional hearing aids. This may at various hearing aid algorithms lead to artifacts and misconduct such. B .:<ol><li>a) In the automatic gain control (AGC: Automatic Gain Control) automatically reduces the gain at high sound levels. the sound changed repeatedly by leaps and bounds, so the gain would be correspondingly varied widely. This means that, for example, ambient or microphone noise is amplified differently depending on Nutzschallpegel what is perceived by the hearing aid wearer as a pumping effect. To avoid these effects, the pumping AGC decay, ie the time or time constant for tracking the gain, typically chosen to be relatively large. But this means that in a conversation with a quieter conversation partner the relatively loud own voice (as measured by the hearing aid!) causes the AGC in transmission phases at low gain values. Speaks the interlocutor for immediately after the hearing aid user has stopped speaking, the AGC is in the decay phase and the gain is correspondingly low. This means that not increase the gain at the usually quieter voice signals of the caller quickly enough so that möglicherwei se the first syllables or words due to lack of reinforcement not be understood.</li><li>b) The approach of an "intelligent directional microphone", which only activates when a voice source of the 0 ° -Vornerichtung is detected, fails because that one's voice as 0 ° source is detected and the directional microphone unfavorably even with a lateral interlocutor is activated.</li><li>c) blind source separation algorithms (BSS: Blind Source Separation) try to present in the microphone signals overlays from the useful sound and various interference signals to separate statistical methods. Again, your voice is identified as a separate source, enabling the extraction of the actual useful signal, which also is usually a speech signal interferes.</li></ol>

p0003The publication <patcit id="pcit0001" dnum="EP1251714A1"><text>EP 1251714 A1</text></patcit> is a digital hearing aid system is known in which an occlusion sub-system compensates for a gain of the hearing aid user's own language in the ear canal. In this case, one of a rear microphone received unwanted signal is fed back and subtracted from the useful signal.

p0004Furthermore, from document <patcit id="pcit0002" dnum="US6041129A"><text>US 6041129 A</text></patcit> discloses a hearing aid, in which the own language of the hearing aid user is amplified or attenuated. Here, the transmitted sound by bone conduction is detected by an accelerometer or motion sensor.

p0005The publication <patcit id="pcit0003" dnum="WO03073790A1"><text>WO 03/073790 A1</text></patcit> is known an apparatus for the detection and discrimination of speech. It is inserted into the ear canal and has an outer microphone for detecting an ambient sound and an internal microphone for detecting sound in the ear canal, which includes its own language. This voice commands can be added to the device.

p0006Further disclosed the document <patcit id="pcit0004" dnum="JP2003304599A"><text>JP 2003304599A</text></patcit> an adjustment method for a hearing aid. The amplification factor of the hearing aid is adjusted in response to an external sound level, and in response to an internal sound in the ear canal.

p0007Furthermore, the publication <patcit id="pcit0005" dnum="WO02085063A2"><text>WO 02/085063 A2</text></patcit> a fitting procedure for a hearing aid. Here an occlusion should be eliminated. To this end, the compression ratio is increased, for example in a language situation, whereby the perceived occlusion is reduced.

p0008In the document <patcit id="pcit0006" dnum="EP1523219A2"><text>EP 1523219 A2</text></patcit> describes a process for retraining and operating a hearing aid. Here, the directional microphone is directed to that direction from which the maximum desired sound, possibly their own language comes.

p0009The document <patcit id="pcit0007" dnum="US20040202333A1"><text>US 2004/0202333 A1</text></patcit> discloses a hearing aid with self-diagnosis. The starting point is also the occlusion, in which their own language is amplified in the ear canal. A canal microphone picks up the unwanted sound and is thus generates a corresponding compensation signal.

p0010Finally, the patent application describes <patcit id="pcit0008" dnum="US6526148B1"><text>US 6526148 B1</text></patcit> a method for blind source separation for speech recognition purposes. It is calculated the internal delay between the microphone signals and used for subtraction of the microphone signals.

p0011The document <patcit id="pcit0009" dnum="DE3325031C2"><text>DE 33 25 031 C2</text></patcit> describes an infrared headset with two microphones. Your signals are in opposite phase to an amplifier fed so that the transfer of own voice is avoided or suppressed.

p0012Further, the patent shows <patcit id="pcit0010" dnum="DE10332119B3"><text>DE 103 32 119 B3</text></patcit> a worn in the ear hearing aid device with a second microphone and a second receiver, which are arranged in a ventilation channel. The signal of the second receiver is inverted in order to avoid that the auditory sound is fed directly.

p0013The object of the present invention is therefore to improve the automatic control of hearing devices at presence of the own voice of the user.

p0014According to the invention this object is achieved by a hearing apparatus according to claim 1. In addition to "normal" sound microphone in the ear canal also connected to the hearing aid housing vibration microphone can be used (eg. B. glued from the inside), which receives the own voice over body sound line preference.

p0015Furthermore, the present invention provides a method of claim 5.

p0016The advantage of the invention is that the activity of one's own voice is detected continuously and very rapidly through the detection approach described above and this information can then be incorporated directly into the control algorithms of the hearing. This triggered by your voice artifacts and mismanagement are avoided.

p0017Preferably, the inherent voice detection means has a level analysis unit, with the respective levels of the two microphone signals on the basis of the level comparison, the presence of the own voice of the user is comparable and can be fixed in the microphone signals. It can be advantageously the occlusion effect of sound advantage in the ear canal, which own voice through the structure-borne noise generated in front of the ear in the ear canal a significantly higher noise level.

p0018Conveniently, only frequencies below 1 kHz are taken into account at the level of analysis. At the low frequencies namely the occlusion effect is most pronounced.

p0019The hearing apparatus according to the invention, a BSS apparatus with which from the / the microphone signal (s), separate sources can be identified, and a signal processing device which can be controlled by the BSS device having, wherein the control of the signal processing device by the BSS device temporarily is frozen when the own voice of the user is detected. Thus, the extraction of the actual useful signal is not disturbed by your voice.

p0020In addition, a hearing device according to the invention may comprise an AGC device for automatic gain adjustment, the temporarily deactivated or their settling in the detection of the own voice of the user is temporarily shortened in the detection of the own voice of the user. Thus, in particular disorders in the conversation can be avoided with a little party.

p0021According to a further embodiment, the hearing device may comprise a directional microphone that is disabled upon detection of the own voice of the user. In this manner, a can "intelligent directional microphone" operate without interference even if the hearing aid wearer himself speaks.

p0022The present invention will now be explained in more detail with reference to the accompanying drawing showing a schematic circuit diagram of a hearing aid according to the invention.

p0023The detail below described embodiment is a preferred embodiment of the present invention.

p0024The problems with the AGC, the BSS and the intelligent directional microphone, which occur when the hearing aid wearer himself speaks, be achieved by the detection of their own voice using a separate ear canal microphone or internal auditory canal microphone MI. This is according to the sketchy character as well as the handset of the chosen here hearing aid in the ear canal GG. In the present example, two external microphones ME1 and ME2 are outside of the ear canal GG for receiving the ambient sound from the environment of the user or the hearing aid wearer.

p0025The detection of own voice is based on the permanent comparison of the external of the hearing aid microphones ME1 and ME2 and the internal auditory canal microphone MI received signals. In this case, this takes place a level analysis PA of the microphone signals. Subsequent to the level analysis PA own voice detection ED provides a simplest case binary signal whether the own voice of the user is detected. Depending on a signal generator SG generates a control signal to control a signal processing unit of the hearing device.

p0026. In the present case the hearing following signal processing units has: a microphone array processing unit MV, eg BSS (adaptive directional microphone), for receiving the microphone signals of external microphones ME1 and ME2, subsequent feedback suppression device RU, subsequent noise reduction unit RR and finally a AGC unit for generating an amplified signal for the listener H.

p0027Both the microphone processor MV including BSS and intelligent directional microphone and the amplification unit AGC are due to the inherent voice detection PA, ED, SG controlled or influenced.

p0028This means that the information on the activity of their own voice can be used directly for controlling the above-mentioned algorithms. For example, as a "freezing" of the BSS-adaptation control is possible when one's own voice is detected. Further, but also a "freezing" of the AGC or a temporary shortening of the decay time is possible when one's own voice is active. Furthermore, can be disabled for the realization of an "intelligent directional microphone" the directional microphone for detection of own voice. This would otherwise be of a 0 ° signal indistinguishable and the directional microphone would be activated.

p0029For the detection of own voice in the present example, a level analysis is performed. These can be combined optionally with a maturity analysis or other analysis.

p0030All external signals appear quieter than the external microphone ME1, ME2, due to the damping effect of the ear device or of the hearing in the case of ITE in the ear canal GG. The well-known for the case hearing aid gain should be included in this level comparison. The level of your voice is at the ear canal microphone due to the direct exposure via bone conduction line in the closed ear canal volume (occlusion) significantly higher than that measured with the external hearing aid microphone ME1, ME2. This level analysis should relate to the frequency below 1 kHz, since this is the occlusion effect is greatest.

p0031The present invention may also be used for headsets and other mobile listening devices.