EP0875107A1

EP0875107A1 - Coding process for inserting an inaudible data signal into an audio signal, decoding process, coder and decoder

Info

Publication number: EP0875107A1
Application number: EP97902223A
Authority: EP
Inventors: Albert Heuberger; Heinz GERHÄUSER; Rainer Perthold; Ernst Eberlein; Roland Plankenbühler; Hartmut Schott; Christian Neubauer
Original assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Current assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Priority date: 1996-03-07
Filing date: 1997-01-24
Publication date: 1998-11-04
Anticipated expiration: 2017-01-24
Also published as: WO1997033391A1; US6584138B1; EP0875107B1; ATE184140T1

Abstract

In a coding method and a coder for introducing a non-audible data signal into an audio signal, the audio signal is first transformed to a spectral range and the masking threshold of the audio signal is determined. A pseudo-noise signal and a data signal are provided and multiplied with each other so a to provide a frequency-spread data signal. The spread data signal is weighted with the masking threshold, and thereafter the audio signal and the weighted data signal are superimposed. In a method and a decoder for decoding a data signal introduced into an audio signal in non-audible manner, the audio signal is first sampled and thereafter the sampled audio signal is filtered in non-recursive manner. The filtered audio signal is subsequently compared to a threshold value so as to retrieve the data signal.

Description

Coding method for introducing an inaudible data signal into an audio signal, decoding method, encoder and decoder

description

The present invention relates to a coding method for introducing an inaudible data signal into an audio signal, to a method for decoding a data signal not inaudibly contained in an audio signal, to an encoder and a decoder.

The transmission of inaudible data signals in an audio signal is used, for example, in range research for radio. The range research serves to reliably determine the audience distribution of individual radio stations. Various methods are known in the prior art for determining the audience distribution of individual radio stations.

A first method works in such a way that the ambient noise is recorded by means of a microphone carried by a listener and compared by means of a reference receiver. The reception frequency of the radio receiver can then be determined from the comparison.

In a second method, the ambient noises are recorded in a compressed form with the information of the exact time in a memory and are then transmitted to a control center. There, the data from high-performance computers is compared with program examples that were recorded during a predetermined period of time, for example one day. In this way, the station heard can be determined.

The methods described above have the following disadvantages. The system described first is not applicable to multi-band reception, multi-standard reception or multi-media reception, since it is only limited to the transmission of frequency-modulated signals. An additional local radiation of other media via free FM channels can only be carried out in individual cases due to the diversity of the program sources. Furthermore, according to this method, the same reception strength is required as that of the receiver of the listener. This condition cannot be achieved in a good reception system or in a car, for example. Another disadvantage is the response time for tuning the reference receiver and the correlation, since this increases with the range of programs and is in the range of minutes. The power consumption of such a method is considerable due to the components used, the receiver, the signal processing, etc. Furthermore, the receiver cannot be designed to be as economical as desired, since the large signal strength is determined directly by the power consumption of the reference receiver. Yet another disadvantage is that only the frequency of the received signal can be determined by the comparison principle, the frequency assignment however depending on the current location. It is therefore necessary to obtain information regarding the location of the listener, for example about the current station tables.

The second method described above has the disadvantage of a considerable memory requirement, since a net data volume of approximately 150 MB results when recording over 24 hours. Even with good compression by a factor of 10, for example, about 15 MB of data is generated every day. As a result, the memories to be used are large and therefore expensive and also have a high current consumption. Furthermore, the determination of the reference programs is difficult since they have to be carried out locally across the country. Another problem is the problem of data protection, since the audio information is obtained directly from the test person's environment. collects and transported to a central evaluation.

In order to avoid the problems described above, several methods have already been proposed in the prior art, in which an identification signal of a transmitter in the form of a data signal is introduced into the audio signal to be transmitted. In this case, the data signal to be transmitted cannot be heard by the listener.

Such methods are described, for example, in WO 94/11989, GB 2260246 A, GB 2292506 A and in WO 95/04430. The disadvantage of this method is that it cannot be ensured that the data signal cannot be heard by the listener at any time during the transmission of the audio signal.

US-A-5,450,490 describes an apparatus and a method for including codes in audio signals and for decoding the same. This system uses different symbols which are encoded by means of entangled frequency lines. In order to ensure that the transmitted data signals cannot be heard at all times, a masking assessment is carried out with regard to the individual frequencies from which the symbols to be transmitted are composed. The disadvantage of this method is that the generation of signals to be transmitted is very complex.

Starting from this prior art, the present invention is based on the object of providing a method for coding and decoding a data signal which is not audibly contained in an audio signal and which ensures that the data signal to be transmitted is not perceived by the human ear is insensitive to interference phenomena and forms a good channel utilization, the data signal being able to be decoded safely and easily. This object is achieved by a coding method according to claim 1 and by a method for decoding according to claim 15.

Starting from this prior art, the present invention is also based on the object of providing an encoder and a decoder for inserting and extracting a data signal which is not audibly contained in an audio signal and which ensures that the data signal to be transmitted is from is not perceived by the human ear, is insensitive to interference and forms a good channel utilization, the data signal being able to be decoded safely and easily.

This object is achieved by an encoder according to claim 18 and by a decoder according to claim 32.

The present invention provides a coding method for introducing an inaudible data signal into an audio signal, comprising the following steps:

a) converting the audio signal into the spectral range;

b) determining the masking threshold of the audio signal;

c) providing a pseudo noise signal;

d) providing the data signal;

e) multiplying the pseudo noise signal by the data signal in order to create a data signal which is spread in terms of frequency;

f) weighting of the spread data signal with the masking threshold; and

g) superimposing the audio signal and the weighted sig- nals,

The present invention provides a method for decoding a data signal which is not audibly contained in an audio signal, with the following steps:

a) sampling the audio signal;

b) non-recursively filtering the sampled audio signal; and

c) comparing the filtered audio signal to a threshold to recover the data signal.

An advantage of the method according to the invention is that information is introduced into an audio signal without being perceived by the human ear, but is reliably decoded by a detector. Another advantage of the present invention is that spread spectrum modulation is used, in which the information or the data signal is spread over the entire transmission band, thereby reducing the susceptibility to interference phenomena and the multipath propagation . At the same time, there is good channel utilization.

According to the present invention, the inaudibility is achieved in that the audio signal, which is for example a music signal to which the data signal or the information is to be added, is subjected to a psychoacoustic calculation. The masking threshold is determined from this and the spread spectrum signal is weighted with it. This ensures that at no point in time is more energy used for data transmission than is permitted psychoacoustically.

According to a preferred embodiment of the present invention, the method for decoding the encoded data signal uses a non-recursive filter (matched filter). ter). The advantage is that this filter can be used for correlation and reconstruction, so that the method for decoding is particularly simple, which is advantageous with regard to a later hardware implementation. A decoder which carries out the method according to the invention can be provided, for example, in the form of a wristwatch, which can easily be worn by test persons.

The present invention also provides an encoder for introducing an inaudible data signal into an audio signal, the

- converts the audio signal into the spectral range;

- determines the masking threshold of the audio signal;

- provides a pseudo noise signal;

- provides a data signal;

- The pseudo noise signal multiplied by the data signal to create a frequency-spread data signal;

- The spread data signal is weighted with the masking threshold; and

- Weighted the audio signal and the weighted data signal.

The present invention provides a decoder for extracting a data signal which is not audibly contained in an audio signal, the

- samples the audio signal;

- filters the sampled audio signal non-recursively; and

- the filtered audio signal with a threshold value equals to recover the data signal.

An advantage of the encoder and decoder according to the invention is that information is introduced into an audio signal without being perceived by the human ear, but is reliably decoded by a detector. A further advantage of the present invention is that spread spectrum modulation is used, in which the information or the data signal is spread over the entire transmission band, thereby reducing the susceptibility to interference phenomena and the multipath propagation. At the same time, there is good channel utilization.

According to a preferred embodiment of the present invention, the decoder uses a non-recursive filter (matched filter). The advantage is that this filter can be used for correlation and reconstruction, so that the method for decoding is particularly simple, which is advantageous with regard to a later hardware implementation. A decoder according to the invention can be provided, for example, in the form of a wristwatch, which can easily be worn by test persons.

Preferred developments of the method according to the invention are defined in the subclaims. Preferred exemplary embodiments of the present invention are explained in more detail below with reference to the accompanying drawings. Show it:

1 shows an embodiment of an encoder according to the invention;

Fig. 2 is an illustration of the transmission frame used to transmit the useful signal;

Figure 3 is a block diagram of the source coding block shown in Figure 1;

4 shows an exemplary embodiment of a decoder according to the invention

FIG. 5 shows a block diagram of the data decoder shown in FIG. 4;

6 shows an exemplary embodiment of a system for determining the audience distribution of a radio station, which uses the methods according to the invention for coding and decoding;

7 shows an exemplary embodiment of a system for determining the audience distribution of a radio station, which uses the methods according to the invention for coding and decoding;

8 shows an exemplary embodiment of a system for identifying audio signals with a unique identification number for identifying sound carriers; and

Fig. 9 shows an embodiment of a system for

Remote control of audio devices, which uses the inventive method for coding and decoding. An exemplary embodiment of an encoder is described in more detail below with reference to FIG. It is pointed out that the circuit shown in FIG. 1 merely represents a preferred exemplary embodiment, and the present invention is not restricted to this.

The coding circuit shown in FIG. 1 consists of a transformation block 100, a psychoacoustic block 102, a data signal generator 104, a source coding block 105, a pseudo-noise signal generator 106, a BPSK baseband modulator 108 (BPSK = Binary Phase Shift Keying = bi¬ nary phase shift keying), a BPSK modulator 110, a device for weighting two signals 112, a reverse transformation block 114 and a superposition or superimposition device 116. In the exemplary embodiment illustrated in FIG. 1, the BPSK baseband modulator 108 is the BPSK modulator 110 and the device for weighting two signals 112 are each formed by a multiplier. A further transformation block 118 is also provided, which transforms the output signal s (l) of the BPSK modulator 110 into the spectral range.

Transformation block 100 is connected to an input ON of the circuit. The output of transformation block 100 is connected to psychoacoustic block 102. The input of the circuit is also connected to an input of the superposition device 116.

The output of the pseudo-noise signal generator 106 is connected to an input of the BPSK baseband modulator 108 and the output of the data signal generator 104 is connected to the input of the source coding block 105, the output of which is in turn connected to the other input of the BPSK baseband modulator 108. The output of the BPSK baseband modulator 108 is connected to an input of the BPSK modulator 110, the other input of which is connected to a signal generator (not shown) which has a cosine shape Applies signal to the other input of the BPSK modulator 110. The output of the BPSK modulator 110 is connected to the further transformation block 118, the output of which is connected to the weighting device 112.

The output of the psychoacoustic block 102 is also connected to the weighting device 112. The output of the weighting device 112 is connected to an input of the reverse transformation block 114. The output of the re-transformation block 114 is connected to a further input of the superposition device 116, the output of the superposition device 116 being connected to an output OFF of the circuit.

A preferred exemplary embodiment of the coding method according to the invention is described in more detail below with reference to FIG. 1.

First, a music signal n (k) is fed in at the "ON" input, which is present, for example, as a digital PCM music signal (PCM = Pulsed Code Modulation). In the transformation block 100, the music signal is first subjected to a windowing with a Hanning window and then converted into the spectral range by means of a fast Fourier transformation (FFT - fast fourier transformation) with a length of 1024 with 50% overlap. Then there is the spectrum N (..) of the music signal n (k) with 512 frequency lines, which is used as an input signal for the psychoacoustics 102. The spectrum of the music signal is simultaneously applied to the superposition device 116, as is shown by the arrow 120.

The spectrum N ((Λ) is divided into critical bands (critical bands) in psychoacoustic block 102. These bands have a width of 1/3 bar, which depends on the sampling frequency (in the present example this is 44.1 kHz or 48, for example) kHz) results in a number of bands of approximately 60. The assignment of the frequencies f (Hz) in bands z (bark) is based on the band division that the human ear makes during the hearing process and is, for example, tabulated in the standard ISO / IEC 11172-3. In these critical bands, the band energy is determined by summing the real part and the imaginary part of the spectrum N (..) according to the following equation:

Ei = Re (N (Ui)) ² + Im (N (ω _i )) ²

This energy distribution is now subjected to a spread. For this, the so-called spreading function is calculated for each band, the calculation following the standard ISO / IEC 11172-3 (1993). The spread curves obtained are then folded with the band energies and the course of the excitation is obtained. The masking threshold W (z) for non-tonal audio signals with one reference point per critical band z can be calculated from this, taking into account the degree of masking.

For tonal audio signals, the masking threshold W (z) is to be set considerably lower. Therefore, a measure of the tonality for each frequency line is determined with the aid of a signal prediction. The prediction determines a predicted vector from the two previous FFTs for each line by adding the phase and magnitude difference to the vector of the last FFT line. An error vector is then formed by forming the difference between the predicted vector and the vector actually obtained from the FFT.

By forming the magnitude of the error vector line by line, a measure of the unpredictability of the signal (abbreviation cw = chaos measure) is calculated for each W. From the "cw" value, which can assume values between 0 - "very tonal" - and 1 - "non-tonal", the measure of masking that has to be taken into account when calculating the masking threshold is calculated.

Alternatively, the masking threshold can also be calculated done differently. The spectral lines obtained from the FFT are summarized in critical bands. These bands have a width of 1/3 bar, which, depending on the sampling frequency (in the present example this is 44.1 kHz or 48 kHz, for example), results in a band number of approximately 60 critical bands. The assignment of the frequencies f (Hz) in bands z (bark) is based on the band division that the human ear makes during the hearing process and is listed in a table, for example, in the standard ISO / IEC 11172-3. In these critical bands, the band energy is determined by summing the real part and the imaginary part of the spectrum N (l _* i) according to the following equation:

Ei = Re (N (Oi)) ² + Im (N (Ui)) ²

It is now assumed that only tonal signals are present in the entire band. In this case (worst case), the masking threshold results by a fixed amount under the energy distribution of the music signal. As a maximum measure of masking, e.g. -18dB can be accepted. The advantage of this method is that the calculation is very simple since neither folds nor predictions have to be made. The disadvantage is that you may Energy reserves that supply the music signal to concealment are not used. However, if sufficient processing gain has been provided, this disadvantage does not bother.

W (z) is now converted into W (OK), this conversion being carried out in accordance with the ISO / IEC 11172-3 standard. The course of the masking threshold W (O) is thus present at the output of block 102, and indicates the energy level up to which the signal can be supplied at a point ( _* -s so that this change remains inaudible.

The data signal generator 104 (DSG) provides the useful data signal x (n), which is generally repeated cyclically in order to decode it at any time in a decoder. possible. The data signal has a bandwidth of, for example, 50 Hz. The data at the output of the DSG 104 are in the form of a binary signal and have a low bit rate 1 / T _χ in the range from 1-100 bit / s. The spectrum of this signal must be very narrow-band in comparison to the spectrum of the signal which is output by the PN signal generator 106 with (* i _χ .

In the exemplary embodiment described in FIG. 1, the useful data signals x (n) consist of words with a length of 11 bits. These data words are built into a frame that has a length of between 26 and 29 bits. 2 shows the structure of such a transmission frame in greater detail. The transmission frame 200 comprises four sections 202, 204, 206, 208. The first section is a synchronous word 202 which consists of seven bits (bits 0 to 6) and is formed by the bit sequence 1111110 in the example shown in FIG. 2. The second section 202 is used for error protection and consists of four bits (bits 7 to 10). The third section 206 contains the data word, which is 11 bits long (bits 11 to 21). The fourth section 208 contains a checksum (checksum) of four bits (bits 22 to 25).

The error protection (section 204 in FIG. 2) is implemented by a non-systematic (15, 11) hammingeode. With this block code, all 1-bit errors can be corrected. In the case of multi-bit errors, the data word received is rejected as incorrect. The advantage of this code is that it can be implemented by simple matrix multiplication without great computational effort and is therefore also suitable with regard to the decoding method.

Since the transmission channel works bit-oriented, the transmission frame must be transmitted with an HDLC protocol (HDLC = high-level data link control). This protocol is modified in such a way that a "0" is not only inserted after six consecutive "1" bits, but also after six "0" bits. Bits a "1". This modification is necessary in order to detect and correct phase rotations that can occur on the channel.

The transmission frame 200 is constructed by the source coding block 105 (FIG. 1). The source coding block 105 is shown in detail in FIG. 3.

The data coding generator 105 provides the data coding generator 105 with the data signals. At input 302 of block 105, the data are present as data words with an 11-bit length, as shown in FIG. 3. The transmission frame is now constructed in such a way that the error protection is first implemented in a first block 304 by the (15, 11) Hamming code. The frame is now 15 bits long. The check sum is then added to the frame in a second block 306. The length is then 19 bits. In block 318, the required coding of the transmission frame is carried out by an HDLC encoder, which leads to a length of the frame of 19 to 22 bits. The binary signal present at the output of block 308 is now converted into an antipodal signal. This can e.g. with the assignment 0 -> 1 and 1 -> -1. To complete the frame, the sync word is added to it in block 310. At the output 312 of the source coding block 105 is the transmission frame with a length of 26 to 29 bits, which is fed to the BPSK baseband modulator 108.

The pseudo-noise signal generator 106 (PNSG) provides the spread signal g (l) with the bit rate 1 / T _g . The bandwidth t-) _q - of this signal determines the bandwidth O _{s of} the spread spectrum signal and, in the embodiment shown in FIG. 1, is in the range of 6 kHz. The higher frequencies that a high-quality music signal offers have been disregarded, taking into account the frequency response of the playback devices (eg portable radios). According to one exemplary embodiment, the PNSG 106 is constructed as a feedback shift register and supplies a pseudo-random one Pseudo-noise sequence (PN sequence) of length N. This sequence must be known in the decoder for decoding the signal.

The ratio T _χ / T _n is called the spreading factor and directly determines the signal-to-noise ratio up to which the method still works reliably. According to the exemplary embodiment described here, the spreading factor is 128 and thus the signal-to-noise ratio S / N = 101oglO (T _x / T _n ) = -21 dB.

The present binary signal g (l) of the PNSG 106 is now converted into an antipodal signal. This can e.g. with the assignment 0 -> 1 and 1 -> -1. After this formatting, the signal is processed and fed to the BPSK baseband modulator.

The BPSK baseband modulator 108 is simple when using antipodal signals, since a sample-wise multiplication corresponds to the BPSK modulation. The resulting signal h (l) = g (l) x '(n) has a bandwidth of (O _n ~ 6 kHz. The amplitude values are -1 and 1. Daε signal has the main maximum at 0 Hz, so it is in the baseband.

The baseband signal h (l) is now fed to the BPSK modulator 110. There the baseband signal h (l) is modulated onto a cosine-shaped carrier cos (CJ _τ t). The frequency of the carrier is half the bandwidth of the spreading band signal in the baseband. The first zero of the modulated spectrum thus comes to be at 0 Hz. As a result, the signal can be transmitted on channels whose transmission function attenuates strongly in the range from 0 to 100 Hz, as can be expected in the case of audio transmissions via loudspeakers and microphone.

Alternatively, the modulation can also be carried out by means of suitable coding instead of using a carrier cosine. Due to its special property of being free of mean values, the man chester code are used. Because of its mean value freedom, there is therefore no energy of the spreading band signal even at 0 Hz, which is important for the transferability. The coding rule for the Manchester code is 0 -> 10 and 1 -> 01. The number of bits is doubled.

The time signal s (l), which is present at the output of the BPSK modulator 110, is now transformed into the spectral range by means of a fast Fourier transformation in the transformation block 118, so that S (tO) is present at the output of the block 118.

The spectral profile of the spread useful signal S (O) is now weighted with the profile of the masking threshold W (W) by the weighting block 112, which means that more noise energy is not introduced at any point in the audio spectrum by the spread spectrum signal than the human ear can perceive. With regard to the demodulation of the useful signal, the statically changing course of the energy distribution in the useful signal has only a minor effect, since the method is particularly powerful in this context.

A reverse transformation is then carried out by an inverse fast Fourier transformation in block 114, so that the encoded music signal is again in the time domain. The 50% overlap must be taken into account for the reverse transformation.

At block 116, the psychoacoustically weighted useful signal is added to the music signal n (k) in the time domain.

At the "OFF" output, the encoder supplies a digital PCM signal n _c (k) which can be transmitted on any transmission link as long as it has a bandwidth of at least 6 kHz.

As an alternative to the exemplary embodiment described above Instead of the input of the circuit, the output of the transformation block 100 can additionally be connected to the superimposition device 116. In this case, the spectral spreading signal and the spectral audio signal are superimposed and then the transformation back into the time domain.

A preferred exemplary embodiment of a decoding circuit is described below, which is used to implement a preferred exemplary embodiment of the method according to the invention for decoding a data signal which is not audibly contained in an audio signal.

The decoder comprises a microphone 400 which receives a music signal emitted, for example, by a radio receiver. The output of the microphone 400 is connected to the input of a low pass 402, the output of which is connected to an amplifier 404 with automatic gain control. The output of amplifier 404 is connected to an analog / digital converter 406. The output of the analog / digital converter 406 is connected to the input of a non-recursive filter 408 (matched FIR filter), the output of which is connected to an input of a bit synchronization control block 410. The output of block 410 is connected to the input of a data decoder 412. The decoded data signal is present at the output of the data decoder 412.

An exemplary embodiment of the decoder according to the invention is described below with reference to FIG. 4. The music signal n _c (k) emitted by the radio receiver is converted into electrical signals by the microphone 400 and fed to the low-pass filter 402. The cut-off frequency of the low-pass filter 402 is dimensioned such that the frequency components in which no data are modulated in are greatly attenuated. In the present embodiment, the cutoff frequency is 6 kHz. The low-pass filtering serves to avoid overfolding which is caused by the later scanning of the signal gnals can arise.

The amplifier 404 with automatic gain control (AGC = Automatic Gain Control) ensures a constant instantaneous power of the input signal in front of the A / D converter 406. This is necessary in order to be able to compensate for temporary damping caused by the channel. It is pointed out that the decoder can be implemented both in terms of hardware and in terms of software. In the case of a software implementation, the amplifier 404 can be omitted.

The A / D converter samples and digitizes the signal.

The matched filter 408 consists of an FIR filter or a non-recursive filter. The filter 408 contains, as coefficients, the reverse sequence of the PN sequence of the transmitter. The PN sequence of the pseudo-noise signal can, for example, be man-coded. In this case, the filter 408 contains, as coefficients, the reverse, man-coded sequence of the PN sequence of the transmitter. Thus, at maximum correlation, the filter 408 generates a peak at the output, the sign of which corresponds to the transmitted symbol. The filter output therefore delivers peaks at a distance of 2 * N in length from the PN sequence, which represent the transmitted data. Since the peaks cannot be clearly determined at all times, the bit synchronization control block 410 is connected downstream of the filter 408.

The synchronization control in block 410 searches for peaks in the output signal of the filter 408, which peaks clearly stand out from the noise cause. If such a peak is found, the output of the filter 408 is scanned in synchronism with the length of the PN sequence in order to recover the transmitted symbols. If a clear peak appears during this time, the sampling time is corrected accordingly. The output of block 410 provides a bit stream that is processed in subsequent data decoder 412. In the event that there is no validly coded signal at the input of the microphone 402, this bit stream represents a random sequence of bits. If the decoder is bit-synchronized, the bit stream contains the transmitted data.

The data decoder 412 decodes the useful data signal from the bit stream from block 410. The data decoder is described in more detail below with reference to FIG. 5. The data decoder 412 includes an IN input connected to a frame synchronization block 502 and an HDLC decoding block 504. Block 502 outputs a trigger signal to block 504. The output of block 504 is connected to the input of a Hamming error correction block 506, the output of which is connected to the input of a check sum block 508. Block 508 is followed by a Hamming data calculation in block 410. The output of block 410 is connected to the output OFF of the data decoder 412, at whose output the data word with a length of 11 bits is present.

The frame synchronization block 502 receives the input bit stream and searches for the synchronization word 202 in it. If it is found, the HDLC decoder 504 is triggered and the input data is decoded accordingly. Then the syndrome is calculated and the error is corrected by the Hammingeode. The checksum is calculated using the bit error-corrected 15-bit word and compared with the transmitted bits. If all of these operations are successful, the 15 bits are decoded with the Hammingeode and the 11 transmitted data bits are output from the decoder.

It is pointed out that the methods for coding and decoding described above are only preferred exemplary embodiments of the present invention, to which the invention is not restricted. The essential features of the coding method according to the invention for introducing an inaudible data signal into an audio signal are converting the audio signal into the spectral range, determining the masking threshold of the audio signal, providing a pseudo-noise signal, providing the data signal, multiplying the pseudo-noise signal with the data signal In order to create a frequency-spread data signal, the weighting of the spread data signal with the masking threshold and the superimposition of the audio signal and the weighted signal.

The essential features of the method according to the invention for decoding a data signal not audibly contained in an audio signal are the sampling of the audio signal, the non-recursive filtering of the sampled audio signal, and the comparison of the filtered audio signal with a threshold value in order to recover the data signal.

A system according to the present invention for determining the listener distribution of individual radio stations on the basis of an identification signal is described in more detail below with reference to FIG. 6. The system described with reference to FIG. 6 uses the coding method described above for introducing the identification signal into the transmitted audio signal and uses the decoding method described above for decoding the signal from the received audio signal.

The system described with reference to FIG. 6 makes it possible to reliably determine the audience distribution of the individual radio stations. The system is independent of the receiving devices used, so that the different listening habits can be taken into account.

The radio transmission can also take place via different media: FM (analog)

Cable (analog and digital)

DAB (220 MHz terrestrial; 1.5 GHz terrestrial and satellite-based)

ADR

Analogue satellite subcarriers (television satellites)

LW / MW / KW

TV sound

It is country-specific which media are relevant for an evaluation, but the system shown in FIG. 6 enables the media listed above to be supported. The detection of the listener range takes place at a predetermined time interval, which can be set depending on the individual case. In one example, the time interval can be 10 seconds. It must also be determined how up-to-date the evaluation has to be. According to the example of a system shown in FIG. 6, the receiver data are acquired overnight. In other exemplary embodiments, it may be sufficient to send in the recording device every 4 weeks for data evaluation.

The system, as shown in more detail in FIG. 6, comprises a recording device which the listener achieves a high level of acceptance in order to ensure the reliability of the data collection. In order to ensure as comprehensive a data acquisition as possible, the recording device is worn on the body of the test receiver or test person, and it is a small device with sufficient battery supply, such as, for example, by rechargeable batteries, which is attractive in design and is easy to use. The batteries are recharged in a charging or docking station. The system according to the invention is provided with the reference number 600 in its entirety in FIG. 6. System 600 consists of the following components. An audio signal is generated in a radio station 602 and an identifier signal is applied to it by means of an identifier 604. The identification signal 604 acts on the audio signal using the coding method described above for introducing an inaudible data signal into an audio signal. The audio signal acted upon by the identification signal is passed on to an antenna 606, which causes radiation 608 of the audio signal. A broadcast receiver 610 consisting of an antenna 612, a receiver 614 and two loudspeakers 616 receives the radiated audio signal. The audio signal received by the antenna 612 is converted via the receiver 614 and the loudspeakers 616 into an audible audio signal 618, which is received by a detection device 620. In the embodiment shown in FIG. 6, the receiving device 620 is designed in the form of a wristwatch. The detection device 620 is operative to extract the identification signal from the received audio signal 618. This is done by means of the method according to the invention for decoding a data signal which is not audibly contained in an audio signal. The identification signal, which is determined by the receiving device 620, is temporarily stored in the receiving device. A so-called docking station 622 is provided in order to receive the wrist watch 620, for example during the night, in order to cause the stored identification data to be transmitted. The docking station 622 is connected via a line 624 and a corresponding connection point 626, to which a telephone 628 can also be connected, to a communication network 630, which in one embodiment is the telephone network. Via the communication network 630, the data or identification data stored by the receiving device 620 are sent to a control center 632, which has a computer 634 in order to evaluate the received data. The computer 634 is connected to a modem 638 via a line 636 connected, which in turn is connected to the communication network 630 via a line 640 and a further connection device 642.

With the system shown in FIG. 6, it is possible to reliably determine the listener data of selected radio stations on a daily basis, the temporal resolution of the system being in the range of a few seconds. The system can be implemented at low cost due to the low-cost technology.

A system according to the present invention for determining the range of a radio station on the basis of an identification signal is described in more detail below with reference to FIG. 7. The system described with reference to FIG. 7 uses the coding method described above for introducing the identification signal into the transmitted audio signal, and uses the decoding method described above for decoding the signal from the received audio signal.

The system according to the invention is provided in its entirety with the reference numeral 700 in FIG. 7. In system 700, an audio signal is generated in a radio station 702, for example in a studio 704, and an identification signal is applied to it by means of an identifier or encoder 706. The application of the audio signal by the identifier 706 takes place using the coding method described above for introducing an inaudible data signal into an audio signal. The audio signal to which the identification signal is applied is passed on to an antenna 708, which causes the audio signal to be radiated 710. A broadcast receiver 712, for example a test receiver, consisting of an antenna 714 and a receiver device 716 receives the radiated audio signal. The receiver 716 shown in FIG. 7 only serves to receive the audio signal. Since this exemplary embodiment is only concerned with determining the transmitter range, the transmitted audio signal can be omitted. An advantage of this procedure is that not only a limited band range in the audio signal can be used to transmit the data signal to determine the transmitter range. It is possible to use the entire bandwidth of the audio signal sent. As a result, either the decoding security or the amount of data transmitted can be increased.

In the exemplary embodiment shown in FIG. 7, the decoder 718, which executes the method for decoding, is formed by a computer 720, which implements the method in terms of software. As can be seen in FIG. 7, the receiver 716 is operatively connected via a line or a cable 722 to a so-called sound card 724 in the computer in order to enable the computer to process the audio signal. Transmission from receiver 712 to decoder 718 over line 722 is analog. In other words, the received audio signal is fed directly from the receiver 712 into the decoder 718.

The decoder 718 is connected via a line 724 to a modem 728, which in turn is connected via a further line 730 to a corresponding connection point 732. The connection point 732 is connected to a communication network 734, for example a telephone network. Via the communication network 734, the data or identification data acquired from the data signal are sent to a center 736 which has a computer 738 in order to evaluate the received data. The computer 738 is connected via a line 740 to a modem 742, which in turn is connected to the communication network 734.

A system for identifying audio signals is described below with the aid of FIG. 8, which system serves to identify sound carriers and copies of sound carriers on the basis of the identification signal introduced into the audio signal. The advantage is that it makes it possible to easily identify possible pirated copies, since each one Sound carrier is provided with an individual identifier ex works.

8a schematically shows the production of a sound carrier, such as a compact disc "CD", in a press shop 800. The press plant 800 comprises a playback device 802 in which a master tape runs, which contains the audio signals to be applied to a CD. The CD is pressed in an 804 press. An encoder 806 is arranged between the press and 804 and the playback device 802. Each CD assigns an identifier signal to the CD, which is introduced into the audio signal. The coding is carried out according to the coding method described above. In order to ensure the generation of individual identification signals for individual CDs, the encoder 806 is assigned a counter which, for example, provides continuous identification numbers as identification signals which are introduced into the audio signal.

The mode of operation of the identifiers on individual CDs is explained in more detail with reference to FIG. 8b. A CD 808, which is provided with an individual identifier, is copied several times, as indicated by the schematically represented playback devices 810. The copies can be made both analog and digital.

After the identifier is built into the audio signal, it is retained even when the audio signal is transmitted in the form of a sound file via the Internet, as is indicated in FIG. 8 by reference number 812. In this way, conclusions can be drawn about the sound file on the sound carrier.

A further exemplary embodiment is described below with reference to FIG. 9. FIG. 9 shows a system for remote control of audio devices which uses the methods according to the invention for coding and decoding. The system according to the invention is provided with the reference number 900 in its entirety in FIG. 9. In system 900, an audio signal is generated in a radio station 902, for example in a studio 904. A data signal or control signal is introduced into the audio signal by means of an encoder 706. The encoder 906 applies the audio signal using the coding method described above for introducing an inaudible data signal into an audio signal. The audio signal to which the signal is applied is passed on to an antenna 908, which causes the audio signal to be radiated 910. A receiver 912, consisting of an antenna 914 and a receiver device 916, receives the radiated audio signal. A decoder is provided in the receiver 916 which extracts the data signal contained in the audio signal in accordance with the decoding method described above. The receiver is constructed in such a way that it responds to the data signal in order to start, for example, the recording of a music program of a radio transmitter. On the basis of the data signal extracted from the audio signal, the receiver causes a recording device 918 to be activated, with which the transmitted audio signal is recorded. This creates a system for radios that provides a method that is comparable to the "VPS" method in television.

According to a further exemplary embodiment of the present invention, a system is created which provides a data channel which operates in parallel with the audio signal in audio devices which process digital data. This data channel has a low bit rate in which information is introduced in accordance with the method described above and extracted in accordance with the decoding method described above.

It is pointed out that the encoder and decoder described above are only preferred exemplary embodiments. The essential features of the encoder for introducing an inaudible data signal The conversion of the audio signal into the spectral range, the determination of the masking threshold of the audio signal, the provision of a pseudo-noise signal, the provision of the data signal, the multiplication of the pseudo-noise signal by the data signal in order to create a frequency-spread data signal are all converted into an audio signal Weighting the spread data signal with the masking threshold and superimposing the audio signal and the weighted signal.

The essential features of the decoder for extracting the data signal contained inaudibly in an audio signal are sampling the audio signal, non-recursively filtering the sampled audio signal, and comparing the filtered audio signal with a threshold value in order to recover the data signal.

Claims

claims

1. Coding method for introducing an inaudible data signal (x (n)) into an audio signal (n (k)), with the following steps:

a) converting the audio signal (n (k)) into the spectral range;

b) determining the masking threshold (W (Cύ)) of the audio signal;

c) providing a pseudo noise signal;

d) providing the data signal;

f) weighting of the spread data signal with the masking threshold; and

g) superimposing the audio signal and the weighted data signal.

2. The coding method according to claim 1, wherein step a) includes applying a fast Fourier transform to the audio signal.

3. Coding method according to claim 1 or 2, wherein step b) comprises the following steps:

bl) splitting the spectrum of the audio signal into critical bands (z); b2) determining the energy in each critical band;

b3) calculating the spreading function for each critical band;

b4) folding the spread curves of all critical bands with the band energies in order to obtain the course of the excitation;

b5) determining the unpredictability of the signal;

b6) folding the unpredictability with the spreading function to obtain a measure of the tonality;

b7) calculating the concealment measure from the tonality; and

b8) calculating the masking threshold from the excitation taking into account the extent of masking.

4. Coding method according to claim 1 or 2, wherein step b) comprises the following steps:

bl) splitting the spectrum of the audio signal into critical bands (z);

b2) determining the energy in each critical band; and

b3) determining the masking threshold from the band energies, taking into account the masking dimension for tonal masking.

5. Coding method according to one of claims 1 to 4, in which the pseudo noise signal has a bandwidth of 6 kHz.

6. Coding method according to one of claims 1 to 5, in which the data signal has a bandwidth of 50 Hz.

7. Coding method according to one of claims 1 to 6, in which the data signal is a channel code a block code.

8. Coding method according to one of claims 1 to 7, in which the pseudo noise signal and daε data signal are converted into antipodal signals before step e).

9. Coding method according to one of claims 1 to 8, in which step e) comprises the following steps:

el) BPSK baseband modulating the data signal with the pseudo noise signal;

e2) BPSK modulating the modulated signal from step el) with a carrier signal whose frequency is in the range of the audible audio spectrum; and

e3) transforming the modulated signal from step e2) into the spectral range.

10. Coding method according to claim 9, in which the carrier signal is cosine-shaped and has a frequency of 3 kHz.

11. Coding method according to claim 9, in which step el) is realized by Manchester coding of the pseudo noise signal.

12. Coding method according to one of claims 1 to 11, in which the weighted data signal from step f) is transformed into the time domain before step g).

13. Coding method according to one of claims 1 to 11, in which in step g) the weighted data signal from step f) with the audio signal in the spectral range is superimposed and the superimposed signal is then transformed back into the time domain.

14. Coding method according to claim 12 or 13, in which the back transformation into the time domain is carried out by a fast Fourier transformation.

15. A method for decoding a data signal which is not audibly contained in an audio signal, comprising the following steps:

a) sampling the audio signal;

b) non-recursive filtering of the sampled audio signal; and

c) comparing the filtered audio signal with a threshold value in order to recover the data signal.

16. The method of claim 9, wherein the audio signal is received with a microphone.

17. The method according to any one of claims 9 or 10, in which before the step a) daε audio signal is low-pass filtered and amplified.

18. Encoder for introducing an inaudible data signal (x (n)) into an audio signal (n (k)) which

- converts the audio signal (n (k)) into the spectral range;

- The masking threshold (W (u))) of the audio signal is determined;

- Provides a pseudo noise signal; - Provides a data signal;

multiplying the pseudo-noise signal by the data signal in order to create a frequency-spread data signal;

- the spread data signal is weighted with the masking threshold; and

- The audio signal and the weighted data signal are weighted.

19. Encoder according to claim 18, which converts the audio signal into the spectral range by means of a fast Fourier transformation.

20. Encoder according to claim 18 or 19, in the determination of the masking threshold

- divides the spectrum of the audio signal into critical bands (z);

- determines the energy in each critical band;

- calculates the spreading function for each critical band;

- the spreading curves of all critical bands fold with the band energies in order to maintain the course of the excitation;

- determines the unpredictability of the signal;

- determines the degree of masking from the tonality; and

- The masking threshold is calculated from the excitation taking into account the specific masking dimension.

21. Encoder according to claim 18 or 19, which in determining the masking threshold

- divides the spectrum of the audio signal into critical bands (z);

- determines the energy in each critical band;

- The masking threshold is determined from the band energies, taking into account the masking measure for the tonal masking.

22. Encoder according to one of claims 18 to 21, in which the pseudo noise signal has a bandwidth of 6 kHz.

23. Encoder according to one of claims 18 to 22, in which the data signal has a bandwidth of 50 Hz.

24. Encoder according to one of claims 18 to 23, which channel-codes the data signal using a block code.

25. Encoder according to one of claims 18 to 24, which converts the pseudo-noise signal and the data signal into antipodal signals before multiplying the pseudo-noise signal by the data signal.

26. Encoder according to one of claims 18 to 25, which when multiplying the pseudo-noise signal with the data signal

- BPSK baseband band modulation of the data signal with the pseudo-noise signal;

- BPSK modulation of the modulated signal from the one with a carrier signal, the frequency of which lies in the range of the audible audio spectrum; and - Converts the modulated signal into the spectral range.

27. The encoder of claim 26, wherein the carrier signal is cosine and has a frequency of 3 kHz.

28. Encoder according to claim 26, in which the multiplication of the pseudo-noise signal by the data signal is carried out by a Manscheεter coding of the pseudo-noise signal.

29. Encoder according to one of claims 18 to 25, which transforms the weighted data signal into the time domain before converting the modulated spreading band signal.

30. Encoder according to one of claims 18 to 25, which superimposes the weighted data signal with the audio signal in the spectral range before converting the modulated spreading band signal and then transforms the superimposed signal back into the time domain.

31. Encoder according to claim 29 or 30, which effects the back transformation into the time domain by a fast Fourier transformation.

32. Decoder for extracting a data signal which is not audibly contained in an audio signal, the

- samples the audio signal;

- filters the sampled audio signal non-recursively; and

- compares the filtered audio signal with a threshold value in order to recover the data signal.

33. The decoder of claim 32, which receives the audio signal with a microphone.

34. Decoder according to claim 32 or 33, which low-pass filters and amplifies the audio signal before sampling.

35. Decoder according to one of claims 32 to 34, in the recovery of the data signal

- finds a correlator peak;

- controls the bit synchronization, and

- performs a frame synchronization and a channel decoding.

36. System for determining the audience distribution of individual radio stations based on an identification signal, with an encoder according to one of claims 18 to 31, which introduces the identification signal into the audio signal, and with a decoder according to one of claims 32 to 35, which transmits the identification signal Pulls out audio signal.

37. System for determining the transmitter range of a radio station on the basis of an identification signal, with a coder according to one of claims 18 to 31, which introduces the identification signal into the audio signal, and with a decoder according to one of claims 32 to 35, which pulls the identification signal out of the transmitted audio signal.

38. System for identifying audio signals with a unique identification number for identifying the sources of copies of sound carriers, with an encoder according to one of claims 18 to 31, which introduces the identification number into the audio signal, and with a decoder according to one of claims 32 to 35 which pulls the identification number out of the transmitted audio signal.

39. System for remote control of audio devices using a control signal, with an encoder according to one of claims 18 to 31, which introduces the control signal into the audio signal, and with a decoder according to one of claims 32 to 35, which extracts the control signal from the transmitted audio signal .

40. System for remote control of audio devices based on a control signal, according to claim 38, in which the recording of an audio signal in a recording device is started and / or ended by the control signal.

41. System for providing a data channel operating in parallel with the audio signal and having a low bit rate in digitally processing audio devices, with an encoder according to one of Claims 18 to 31, which introduces the information into the audio signal, and with a decoder according to one of claims 32 to 35, which extracts the information from the transmitted audio signal.