EP2622879B1 - Method and device for frequency compression - Google Patents

Description

The present invention relates to a method for frequency compression of an audio signal in a listening device. Moreover, the present invention relates to a corresponding device for frequency compression. A hearing device is understood to mean any sound-emitting device which can be worn in or on the ear, in particular a hearing aid, 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, one or more microphones 2 for receiving the sound from the environment are installed. A signal processing unit 3, which is also integrated into 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

Many hearing losses can be compensated by using frequency-dependent amplification combined with dynamic compression. However, there are also hearing losses where amplification has no effect or is detrimental. An example of this is hearing loss with so-called "dead regions". "Dead regions" are frequency ranges in which spectral components can no longer be audibly amplified.

One possible technique to deal with the above problem is frequency compression. Here, spectral components from a source frequency range, typically at higher frequencies and in which no gain is to be applied (e.g., "dead region"), are shifted to a lower target frequency range. Audibility is generally guaranteed in this target frequency range, which is why amplification can be usefully applied.

Known frequency compressions work as follows, for example: A compression rule for an individual hearing loss is tailored, wherein the compression rule defines which source frequency should be compressed or mapped to which target frequency. The practical realization of this compression rule is carried out by a filter bank. This means that the compression rule defines which source channel of the filter bank is mapped or compressed to which target channel. The smallest element So this process is a channel. This means that the spectral components within a channel are not compressed. In addition, the possible positions of the channels are defined by the structure of the filter bank and thus fixed (fixed filter bank grid). Such a system is for example from the US 6,577,739 B1 known.

However, the described method for frequency compression is particularly unsuitable for speech sound. In a speech sound voiced sounds is a fundamental frequency and several harmonics, which are found at integer multiples of the fundamental frequency. This is called the fine structure of the signal. The fine structure is responsible for the perception of the pitch of the speech sound. The amplitudes of the fundamental frequency and the harmonics define the color of the sound and form the so-called spectral envelope. For example, the spectral envelope of vowels shows a typical formant structure in each case. The spectral envelope carries the essential information that allows discrimination of the different sounds (e.g., distinguishing the vowels).

As described above, prior art frequency compression is accomplished by shifting source channels on a fixed filter bank grid. The fixed filter bank grid is defined by the filter bank structure and not by the harmonic structure of the signal. Therefore, movement of source channels on the fixed filter bank grid to their destination channels in accordance with the compression rule destroys the harmonic structure. The reason for this is that when moving the harmonic structure is just not considered. That the harmonics no longer inevitably appear at integer multiples of the fundamental frequency after compression. The destruction of the harmonic structure, however, leads to audible artifacts.

In the document EP 2 099 235 A2 is described a system for frequency compression, by means of which a high-frequency Part of a spectral envelope is compressed, while a low-frequency range remains unchanged. In order to avoid artifacts, such as may be caused by the frequency compression in the high-frequency signal component, it is provided in the system to randomize the phase information of the high-frequency component.

In the document US 4,051,331 A For example, a hearing aid system and method are described in which frequency compression of a speech signal is performed by detecting and altering both a formant frequency and a fundamental frequency. Based on the thus obtained changed frequency values, an artificial output signal is generated by means of a tone generator.

In the document WO 2006/133 431 A2 For example, FFT (Fast Fourier Transform) frequency compression is described in which frequency bifurcated frequency bins are multiplied by a complex value to adjust their phase values.

The object of the present invention is therefore to be able to better avoid artifacts in the frequency compression.

According to the invention, this object is achieved by a method for frequency compression of an audio signal in a listening device by obtaining an amplitude information of a source channel from a plurality of frequency channels of the audio signal and impressing an amplitude corresponding to the amplitude information to a signal in a destination channel of the plurality of frequency channels to which the source channel at Frequency compression is mapped, wherein the phase of the signal is maintained in the target channel.

Moreover, according to the present invention, there is provided an apparatus for frequency-compressing an audio signal for a listening device, comprising estimating means for obtaining amplitude information of a source channel of a plurality of frequency channels of the audio signal and processing means for impressing an amplitude corresponding to the amplitude information on a signal in a destination channel of the plurality of frequency channels the source channel for frequency compression is to be mapped, wherein the processing means is adapted to maintain the phase of the signal in the target channel.

Advantageously, the amplitude information in a source channel of an audio signal is separated from the actual signal and used to impose a corresponding amplitude on a signal in a destination channel. Frequencies in the target channel are not affected thereby, whereby the harmonic structure of the audio signal can be maintained.

The amplitude information may be a mean channel amplitude. This is easy to win for a channel and can also be transferred to a target channel with little effort.

The amplitude information is preferably a spectral model of the audio signal, the spectral model is subjected to frequency compression, and the amplitude to be applied to the signal of the destination channel is determined from the compressed spectral model. For example, the spectral model is the spectral envelope resulting from the amplitudes the fundamental frequency and harmonic of a harmonic signal. The spectral model thus represents a function that models the amplitude values over the frequency.

The amplitude to be recorded for the target channel can be obtained by sampling the compressed spectral model. Thus, the amplitude for a certain frequency is obtained from the compressed spectral model or the compressed spectral envelope.

Alternatively, the amplitude to be recorded can be obtained by integrating or summing values of the compressed spectral model in the region of the target channel. As a result, a mean amplitude value for the target channel is determined from the spectral model.

In one embodiment, at least one channel amplitude is obtained for each of the frequency channels and the spectral model of the audio signal is obtained from the channel amplitudes. Thus, at least one value per frequency channel is provided for the spectral model.

The spectral model can be obtained by interpolation (spline). The individual points are connected by linear functions, quadratic functions, cubic functions and the like. The spectral model can also be a polynomial function. In this case, the spectral model or the spectral envelope is simulated by an analytical function. From this in turn, amplitude values can be obtained without high computational effort.

The spectral model can also be obtained by a linear predictive coefficient (LPC) analysis in the time domain. This can be dispensed with a filter bank.

However, if the spectral model is obtained, for example, by interpolation, it is favorable if the device for frequency compression comprises a polyphase filter bank to provide the audio signal in multiple frequency channels. This makes it possible to generate only positive frequency components in the channels.

The device according to the invention is particularly advantageously used in a listening device and in particular in a hearing aid. Thus, a frequency compression in hearing aid users can be realized with fewer artifacts.

FIG 1

den prinzipiellen Aufbau eines Hörgeräts gemäß dem Stand der Technik;

FIG 2

ein Spektralmodell eines Audiosignals vor einer Kompression;

FIG 3

das Spektralmodell von FIG 2 nach der Kompression;

FIG 4

ein harmonisches Signal mit den Amplituden des komprimierten Spektralmodells; und

FIG 5

ein harmonisches Signal, bei dem die Amplituden durch Integralbildung gewonnen werden.

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 spectral model of an audio signal before compression;

FIG. 3

the spectral model of FIG. 2 after compression;

FIG. 4

a harmonic signal with the amplitudes of the compressed spectral model; and

FIG. 5

a harmonic signal in which the amplitudes are obtained by integral formation.

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

The main object of the present invention is to leave the spectral fine structure, in particular of a harmonic signal, untouched by subjecting only the amplitude information of a spectrum to compression. In particular, for example, only the spectral envelope, which represents a measure of the magnitude of the amplitude in the spectrum, is compressed.

In a first implementation variant, the input signal is spectrally decomposed by a filter bank. For each channel participating in the compression process, a corresponding Channel strength calculated. Examples of channel strengths are the amplitude, the amplitude square, or any other measure of the power or strength of the signal in the corresponding channel. The channel strengths can be interpreted as samples of the spectral envelopes that are to be compressed. The channel strength represents an amplitude information in the sense of the present application.

The compression is achieved by shifting the channel strengths from the source channels to the destination channels according to a predetermined compression rule. The original channel strengths of the destination channel (before compression) will be overwritten. That is, according to the present invention, the phase of an original signal (before compression) is maintained in the target channel. Only the channel strengths are modified. Thus, for example, after the filter bank, the envelope is impressed on the respective signals, and the phases are retained.

In principle, the compression rule according to the present invention is similar to the compression rule of a compression system according to the prior art. The difference between the prior art approach and the inventive approach is that, according to the approach of the invention, only the channel strengths are shifted while in the prior art approach the complete channel signals are shifted. In the approach according to the invention, therefore, the spectral fine structure is retained. A harmonic remains a harmonic. Optionally, only its amplitude is varied.

In a second implementation variant, the input signal is spectrally decomposed using a filter bank. The channel strengths of all channels to be compressed are used to obtain a spectral model (eg an envelope). This spectral model is obtained, for example, by linear interpolation, quadratic interpolation, cubic interpolation or by analytical modeling obtained using a polynomial function. The spectral model or the envelope is compressed according to the compression rule. Finally, the compressed spectral model is used to calculate the strengths of the target channels. The phases of the destination channels are not modified as in the first implementation variant described above.

Following concrete embodiments are reproduced in detail.

FIG. 2 shows a spectral model of an input signal of a hearing aid. The channel strength (eg, amplitude, power, etc.) is plotted against frequency f for each of the frequency channels 10. The respective channel strength is symbolized by a point 11. Adjacent points 11 are each connected by a straight line. This results in a spectral envelope 12 by linear spline interpolation. The spectral envelope 12 thus represents a spectral model of the input signal.

Due to a hearing aid's internal compression rule, a high-frequency portion 13 of the entire spectrum is to be compressed. The compression starts at a frequency f_cut_off. The range to be compressed ranges from this frequency f_cut_off to the highest processed frequency channel. The channels in the compression area 13 may be referred to as source channels 14 for frequency compression.

All frequencies above the frequency f_cut_off are thus compressed according to the same, dependent on the hearing loss compression rule. In this compression, the original envelope 12 becomes the compressed envelope 12 'according to FIG FIG. 3 compressed. In a frequency range below the frequency f_cut_off, the two envelopes 12 and 12 'coincide with each other. Above the frequency f_cut_off is the linear interpolated curve of FIG. 2 compressed according to the compression rule on significantly less frequency channels. The structure of the envelope is Although, in terms of amplitude sequence, essentially retained, but the curve was compressed in the frequency direction. The highest frequency f_max 'after compression thus lies below the frequency f_max in the uncompressed case according to FIG. 2 , However, this also means that a plurality of source channels 14 are mapped to fewer destination channels 15. The target channels 15 each have the same width as a source channel 14. The channel structure is therefore unaffected by the compression. From the compressed envelope 12 '(compressed spectral model), the channel strength can thus be determined for each target channel 15, as can be seen from the examples of FIG 4 and FIG. 5 will be shown.

FIG. 4 shows a section of the target channels 15 of FIG. 3 , In the middle between the channel boundaries of each target channel 15, the compressed envelope 12 'is scanned. It can already be seen here that the sampled values are not necessarily at the break points of the compressed envelope 12 '. Thus, it is not the strength of a source channel 14 that maps exactly to the strength of a target channel 15. Rather, the value for the channel strength of the target channel is obtained directly from the sample, which results at the respective channel center of the compressed envelope 12 '. However, according to another embodiment, the sampling may also be performed at a different frequency position within each destination channel 15. Thus, for example, the sampling can also take place at a channel boundary.

According to the example of FIG. 5 the value of a destination channel 15 is determined in another way. Namely, it is determined by averaging based on an integral or a sum of all values of the compressed envelope 12 'within each channel. The respective mean value 16 is then a measure of the strength of the target channel 15. Again, the channel structure and in particular the distance between harmonics of the frequency compression remains unaffected. Only the amplitude of the spectral components in the compressed range is adapted or changed.

According to a modified embodiment, the decomposition of the input signal into the spectral fine structure and the spectral envelope can also be effected by means of an LPC (linear predictive coefficient) analysis and calculation of the residual signal in the time domain. Thus, no filter bank is necessary to obtain the envelope, as is required for the calculation in the frequency domain.

Thus, according to the invention, a decomposition of the input signal into a spectral fine structure and a spectral envelope (spectral model) takes place and the spectral envelope is compressed independently of the spectral fine structure by a compression rule dependent on the hearing loss. The spectral fine structure is retained. Consequently, the harmonic structure of a tonal signal remains unaffected, so that the described artifacts do not occur or are reduced. Frequency estimation is not necessary for this procedure.

Claims (12)

1. Method for frequency compression of an audio signal in a hearing apparatus, by

   - obtaining amplitude information of a source channel (14) from a plurality of frequency channels (10) of the audio signal,

   characterized by

   - applying an amplitude corresponding to the amplitude information to a signal in a target channel (15) of the plurality of frequency channels (10), on which the source channel (14) is mapped during the frequency compression, wherein the phase of the signal in the target channel is maintained.
2. Method according to Claim 1, wherein the amplitude information is a channel intensity, which constitutes a measure for the signal power or signal intensity in the corresponding channel.
3. Method according to Claim 1, wherein the amplitude information is a spectral model (12) of the audio signal, which reproduces amplitude values of a spectral envelope in an exemplary fashion, the spectral model (12) is subjected to the frequency compression in order to form a compressed spectral model and the amplitude to be applied to the signal of the target channel (15) is established from the compressed spectral model (12').
4. Method according to Claim 3, wherein the amplitude to be applied is obtained by scanning the compressed spectral model (12').
5. Method according to Claim 3, wherein the amplitude to be applied is obtained by forming an integral or a sum of amplitude values of the compressed spectral model (12') in the region of the target channel (15).
6. Method according to one of Claims 3 to 5, wherein at least one channel amplitude is obtained for each of the frequency channels (10) and the spectral model (12) of the audio signal is obtained from the channel amplitudes.
7. Method according to Claim 6, wherein the spectral model (12) is obtained by interpolation of the channel amplitudes.
8. Method according to Claim 6 or 7, wherein the spectral model (12) is a polynomial function.
9. Method according to Claim 6, wherein the spectral model (12) is obtained by LPC-analysis in the time domain.
10. Device for frequency compression of an audio signal for a hearing apparatus, comprising

    - an estimation apparatus for obtaining amplitude information of a source channel (14) of a plurality of frequency channels (10) of the audio signal and

    characterized by

    - a processing apparatus for applying an amplitude corresponding to the amplitude information to a signal in a target channel (15) of the plurality of frequency channels (10), on which the source channel (14) is to be mapped for the frequency compression, wherein the processing apparatus is designed to maintain the phase of the signal in the target channel.
11. Device according to Claim 10, which has a polyphase filter bank in order to decompose the audio signal spectrally and, as a result of this, to provide it in a plurality of frequency channels (10).
12. Hearing apparatus with a device according to Claim 10 or 11.

Images