EP2434781A1 - Method for reconstructing a speech signal and hearing aid - Google Patents

Abstract

The method involves storing a preset amplitude spectrum of a speech content, and detecting an amplitude spectrum (15) of an input signal that contains a speech signal. A consistent part (s) and an inconsistent part (n) of the preset amplitude spectrum with the amplitude spectrum of the input signal are detected. Amplification of the input signal in the inconsistent part is changed such that consistency with the preset amplitude spectrum advanced against original amplification is achieved. Processing of the input signal is taken place in different frequency channels. An independent claim is also included for a hearing aid comprising a detection device.

Description

The present invention relates to a method for reconstructing a speech signal. Moreover, the present invention relates to a hearing device with which a speech signal can be reconstructed. The term "hearing device" is here understood to mean any sound-emitting device which can be worn on or in the ear, in particular a hearing device, a headset, headphones and the like.

Hearing aids are portable hearing aids that are used to care for the hearing impaired. In order to meet the numerous individual needs, different types of hearing aids such as behind-the-ear hearing aids (BTE), hearing aid with external receiver (RIC: receiver in the canal) and in-the-ear hearing aids (IDO), e.g. Concha hearing aids or canal hearing aids (ITE, CIC). The hearing aids listed by way of example are worn on the outer ear or in the ear canal. In addition, bone conduction hearing aids, implantable or vibrotactile hearing aids are also available on the market. The stimulation of the damaged hearing takes place either mechanically or electrically.

Hearing aids have in principle as essential components an input transducer, an amplifier and an output transducer. The input transducer is usually a sound receiver, z. As a microphone, and / or an electromagnetic receiver, for. B. an induction coil. The output transducer is usually used as an electroacoustic transducer, z. As miniature speaker, or as an electromechanical transducer, z. B. bone conduction, realized. The amplifier is usually integrated in a signal processing unit. This basic structure is in FIG. 1 shown using the example of a behind-the-ear hearing aid. In a hearing aid housing 1 for carrying behind the ear are one or more microphones 2 to Built-in sound recording from the environment. A signal processing unit 3, which is also integrated in the hearing aid housing 1, processes the microphone signals and amplifies them. The output signal of the signal processing unit 3 is transmitted to a loudspeaker or earpiece 4, which outputs an acoustic signal. The sound is optionally transmitted via a sound tube, which is fixed with an earmold in the ear canal, to the eardrum of the device carrier. The power supply of the hearing device and in particular the signal processing unit 3 is effected by a likewise integrated into the hearing aid housing 1 battery. 5

An essential aspect in the care of the hearing-impaired with hearing aids is speech intelligibility. This means that a word or a word component should also be recognized as such or as such by the hearing aid wearer. An essential role in speech intelligibility is played by the consonants, in particular, for example, the "S". In the listening situation "speech in noise", consonants are often not clearly audible or are heard as other consonants. For example, the word "nonfiction" can be heard as a "textbook".

To improve speech intelligibility, noise reduction algorithms or speech enhancement algorithms are mostly used. In the listening situation "speech in broadband noise", only a directional microphone increases speech intelligibility. Directional microphones work only meaningful if the noise and the language come from different directions. Other noise reduction algorithms, e.g. Wiener filters, do not increase speech intelligibility in noise. At best they reduce the listening effort.

The object of the present invention is thus to propose a method and a hearing device with which increased speech intelligibility can be ensured.

According to the invention, this object is achieved by a method for reconstructing a speech signal by storing a predetermined amplitude spectrum of a speech component, detecting an amplitude spectrum of an input signal containing the speech signal, detecting at least a matching part and a mismatched part of the predetermined amplitude spectrum with the amplitude spectrum of the input signal and Changing a gain of the input signal in the mismatched portion of the amplitude spectrum such that a greater match than the original gain with the predetermined amplitude spectrum is achieved.

In addition, the invention provides a hearing device with which a speech signal is reconstructable, comprising a memory device for storing a predetermined amplitude spectrum of a speech component, a detection device for detecting an amplitude spectrum of an input signal containing the speech signal, a detection device for detecting at least one matching part and mismatched part of the predetermined amplitude spectrum with the amplitude spectrum of the input signal and an amplifying means, with which a gain of the speech signal in the mismatching part of the amplitude spectrum is variable such that a relation to an original gain further match with the predetermined amplitude spectrum is achieved.

Advantageously, the input signal containing the speech signal and any noise is examined for predetermined patterns in the amplitude spectrum. If certain patterns or pattern parts are detected in the amplitude spectrum of the input signal, the remaining amplitude spectrum can be adapted to the given pattern by changing the gain. Thus, for example, a predetermined Language component of an amplitude spectrum are "worked out".

Preferably, the processing of the input signal takes place in a plurality of frequency channels, and each amplitude spectrum is characterized by one amplitude value per frequency channel. This is equivalent if the processing takes place in digital frequency values and an amplitude value is assigned to each frequency value in a specific amplitude spectrum.

It is particularly advantageous if the language component is a consonant. Consonants have a greater importance in speech intelligibility than vowels.

In a further embodiment, a predetermined amplitude spectrum of a plurality of speech components is stored, the amplitude spectrum of the input signal is checked for at least partial coincidence with each of the predetermined amplitude spectra, and the gain is changed in response to the at least partially coincident predetermined amplitude spectrum. Thus, for example, the reconstruction of several different consonants in an input signal can be achieved in a targeted manner if corresponding parts of amplitude spectra are detected.

The detection for matches can be limited to formants. Formants are rapidly detectable in a spectrum and they carry the essential information for the distinctness of language components.

In another embodiment, the gain may be changed to achieve full match with the predetermined amplitude spectrum. In this way, certain speech components can be made very clearly audible.

Detecting at least one matching part and a mismatched part of the predetermined amplitude spectrum with the amplitude spectrum of the input signal may include balancing the absolute values of the predetermined amplitude spectrum with the absolute values of the amplitude spectrum of the input signal. Thus, it is not necessary for the amplitude spectrum of the input signal to be absolutely identical to the stored amplitude spectrum. Rather, a relative agreement of the spectral values is sufficient.

In addition, after changing the gain, the input signal can be additionally amplified in its entirety or transferred to another frequency range. This further increases the audibility of the reconstructed speech component.

The inventive method for reconstructing a speech signal can be used particularly advantageously in signal processing in a hearing aid.

FIG 1

den prinzipiellen Aufbau eines Hörgeräts gemäß dem Stand der Technik;

FIG 2

ein schematisches Zeitsignal eines Konsonanten;

FIG 3

das Spektrum des Zeitsignals von FIG 2;

FIG 4

eine Detektion und Rekonstruktion eines Spektrums in einem ersten Ausführungsbeispiel; und

FIG 5

eine Detektion und Rekonstruktion eines Spektrums in einem zweiten Ausführungsbeispiel.

The present invention will now be explained in more detail with reference to the accompanying drawings, in which:

FIG. 1

the basic structure of a hearing aid according to the prior art;

FIG. 2

a schematic time signal of a consonant;

FIG. 3

the spectrum of the time signal of FIG. 2 ;

FIG. 4

a detection and reconstruction of a spectrum in a first embodiment; and

FIG. 5

a detection and reconstruction of a spectrum in a second embodiment.

The embodiments described in more detail below represent preferred embodiments of the present invention.

When a consonant is pronounced, a corresponding time signal can be obtained, as in FIG. 2 symbolically indicated. From this time signal, a sample or snapshot sa with a certain time width can be obtained.

In a customary manner, a short-term spectrum can be obtained from the temporal snapshot sa, which is described by way of example in FIG FIG. 3 is shown. The short-term spectrum of a consonant has a typical shape. In particular, a consonant can be identified on the basis of its specific position of the formants 10, 11.

For carrying out the method according to the invention or for realizing the hearing device according to the invention, one or more consonants are now recorded in an undisturbed environment. The spectrum of each consonant is, for example, sampled digitally and the individual samples 12 of the short-term spectrum 13 are stored in a memory device of the hearing device and in particular of the hearing device. In this way, for each consonant being recorded, a short-term spectrum can be stored in the hearing device.

During operation, the hearing device now permanently analyzes the input signal and searches for the spectral pattern of the consonant or the patterns of the stored consonants. Usually then the consonant (hereinafter the method is described by means of a single consonant) is spoken against a background noise. In the embodiment of FIG. 4 The background noise has the noise spectrum 14, while the consonant, ie the useful signal, has the spectrum 15. In a section n of the spectrum, the noise spectrum 14 predominates, while in a other section s, the signal spectrum 15, namely that of the spoken consonant predominates. In the area s, when scanning the entire spectrum, the consonant spectrum 15 is actually sampled by the signal processing. The sampled spectrum 15 is compared with the stored spectrum 13. If this part of the spectrum 15 has, for example, a very characteristic form, it can be concluded that the stored consonant has been spoken. It is then assumed that the signal in the spectral range n is superimposed by a noise. The amplification is then lowered, for example channel by channel, so that the stored spectrum 13 of the consonant is also given in the spectral range n. This reduction in gain is in FIG. 4 symbolized by the arrows 16. The spectrum is thus reconstructed in the region n or extrapolated from the measured consonant spectrum 15 using the stored spectrum 13. The resulting spectrum corresponds to that of the stored spectrum, which was recorded without background noise. If the hearing aid wearer now reproduces the reconstructed spectrum, he hears the consonant more clearly, since the background noise was damped. He practically hears the consonant as if it were spoken calmly.

In the example of FIG. 4 Only a very small part of the short-term spectrum (area s) is detected as a prominent area. Whether this small area alone is sufficient to identify the corresponding consonant depends on the performance of the detection device in the hearing device. As a rule, a single peak or formant 10 will not be sufficient to correctly identify a consonant. Therefore be with FIG. 5 explains a second embodiment, in which an identification is easier to carry out. The same consonant spectrum 15 is here superimposed with an interference spectrum 14 'of lower amplitude. Only in a very small area n 'dominates the noise. In the much larger area s' dominates the consonant spectrum 15. In particular, the formants protrude 10 and 11 on the interference spectrum 14 '. On the basis of the formants 10, 11 and optionally additionally based on the spectral characteristics in the areas s', the spoken consonant can be identified more easily by comparison with the stored spectrum 13 than in the case of FIG FIG. 4 , To reconstruct the entire spectrum, only the gain in the region n 'according to the arrows 16' has to be reduced. The reconstructed spectrum then no longer has any interfering components. The hearing aid wearer perceives the spoken consonant as if it were spoken in peace.

The reconstructed consonants can then be further processed, e.g. through specific reinforcement. Likewise, the reconstructed consonants can be pushed, for example, by frequency shift / compression in an audible range for the hearing aid wearer.

The above examples always refer to consonants. The method can also be applied to other language components such as whole words or log atoms.

Analogous to the reduction of noise components and the useful signal components of the speech component can be raised in terms of higher gain. The entire spectrum is then raised uniformly, for example, in the regions s ', while in the region n' it is only channel-specifically raised in such a way that ultimately the pattern of the stored spectrum 13 is formed.

Advantageously, it is possible by the present invention to detect by statistical methods a spectral pattern of a speech component in a background noise. The noisy pattern is then reconstructed from a known pattern by specific reduction of gain (in the relevant channels). Subsequently, the reconstructed language component can be further processed. Overall, the respective language component is thus suppressed, whereby an increased speech intelligibility in noise can be achieved.

Claims (10)

Method for reconstructing a speech signal, characterized by Storing a predetermined amplitude spectrum (13) of a speech component, Detecting an amplitude spectrum (15) of an input signal containing the speech signal, - Detecting at least one matching part (s) and a mismatched part (s) of the predetermined amplitude spectrum (13) with the amplitude spectrum (15) of the input signal and - Changing a gain of the input signal in the mismatching part (s) of the amplitude spectrum (15) such that an over the original gain further agreement with the predetermined amplitude spectrum (13) is achieved. The method of claim 1, wherein the processing of the input signal is carried out in a plurality of frequency channels, and each amplitude spectrum is characterized by a respective amplitude value per frequency channel. The method of claim 1 or 2, wherein the speech component is a consonant. Method according to one of the preceding claims, wherein in each case a predetermined amplitude spectrum (13) of a plurality of speech components is stored, and the amplitude spectrum (15) of the input signal is checked for at least partial agreement with each of the predetermined amplitude spectra (13), and the gain is dependent is changed by the at least partially coincident, predetermined amplitude spectrum (13). Method according to one of the preceding claims, wherein the detection is limited in terms of matches on formants (10, 11). Method according to one of the preceding claims, wherein the gain is changed so that a complete match with the predetermined amplitude spectrum (13) is achieved. A method as claimed in any one of the preceding claims, wherein the detecting includes balancing the absolute values of the predetermined amplitude spectrum (13) with the absolute values of the amplitude spectrum (15) of the input signal. Method according to one of the preceding claims, wherein after changing the amplification the input signal is additionally amplified or transferred into another frequency range. A method for processing a speech signal in a hearing aid with a reconstruction of the speech signal according to one of the methods according to one of the preceding claims. Hearing apparatus with which a speech signal can be reconstructed,
marked by a memory device for storing a predetermined amplitude spectrum (13) of a speech component, a detection means for detecting an amplitude spectrum (15) of an input signal containing the speech signal, - Detection device for detecting at least one matching part (s) and a mismatched part (s) of the predetermined amplitude spectrum (13) with the amplitude spectrum (15) of the input signal and - An amplification device with which a gain of the speech signal in the mismatching part (s) of the amplitude spectrum is variable such that a relation to an original gain on going Agreement with the predetermined amplitude spectrum (13) is reached.

Images