JP6353402B2 - Acoustic digital watermark system, digital watermark embedding apparatus, digital watermark reading apparatus, method and program thereof - Google Patents

Description

  The present invention transmits audio signals such as voice and music (hereinafter also simply referred to as “acoustic signals”) with parallel information such as text added thereto, and reproduces the audio signals with good quality on the receiving side. The present invention relates to an acoustic digital watermark technique for reading added parallel information.

  FIG. 1 is an example of the prior art of an information communication system (hereinafter also referred to as “acoustic digital watermark system”) using an acoustic digital watermark. In the transmission unit 91, the digital watermark embedding unit 92 embeds parallel information as an electronic watermark in the input acoustic signal (hereinafter also referred to as “input acoustic signal”), and outputs a reproduction signal. The reproduction signal is reproduced from the speaker 93. The receiving unit 94 collects the reproduction sound of the reproduction signal via the acoustic space by the microphone 95, and the digital watermark reading unit 96 reads the parallel information from the sound collection signal and also outputs the acoustic signal (hereinafter referred to as “output acoustic signal”). Output). As the speaker 93 of the transmission unit 91, for example, an announcement system in a store or a train, or an acoustic speaker for outdoor advertising or an event can be used. As the microphone 95 of the receiving unit 94, a microphone built in a mobile phone or a smartphone can be used in addition to a vocal microphone or an announcement microphone.

  The audio digital watermarking system of FIG. 1 is, for example, transmitting music titles as text as parallel information while playing music from a speaker, or playing game information as parallel information while playing sports video and sound from a speaker. In addition, it is possible to transmit the URL of a ticket sales site or to transmit a foreign language translation as text as parallel information while reproducing a Japanese announcement from a speaker.

  As a conventional technique of such a digital watermark method, the methods of Non-Patent Document 1 and Non-Patent Document 2 are known.

Matsuoka, Nakajima, Yoshimura “Sound wave information transmission method to acquire information with microphone of mobile terminal”, NTT DoCoMo Technical Journal, 2006, VOL.14 NO.2, pp.6-13 Shigeki, “Development of Digital Watermark Embedding Techniques for Non-contact Extractable Music”, Information Processing Society of Japan Research Report, 2005, 2005-MUS-60, pp.1-6

The important thing about audio watermarking is
(1) Securing necessary communication speed (bit rate) of parallel information (securing communication speed of parallel information).
(2) The embedded parallel information can be reliably read on the receiving side (ensure communication reliability).
(3) When a person listens to the output sound signal, deterioration is not detected or is not anxious compared to the input sound signal before the parallel information is embedded (maintenance of sound quality).
It is.

  In Non-Patent Document 1, a high frequency of an input acoustic signal, for example, a signal of 5 kHz to 10 kHz is replaced with an OFDM signal. In this method, the conditions (1) and (2) are satisfied, but it cannot be said that the condition (3) is satisfied. In Non-Patent Document 1, only the spectral outline is preserved in the high frequency range of the input acoustic signal, but the original acoustic information is lost. Moreover, since the parallel information is embedded only in the high frequency as an OFDM signal, a microphone capable of receiving high frequency with high sensitivity is required.

  In Non-Patent Document 2, when a stereo speaker and a stereo microphone are used, it is said that (1), (2), and (3) can ensure sufficient performance, but when a monaural speaker and a monaural microphone are used. Is equivalent to reproducing one of two predetermined low-frequency signals, for example, 0 to 100 Hz and 100 to 200 Hz, with one of the signals set to zero (as a sound lacking the sound in the band) Assuming that ambient noise exists in the acoustic space between the microphones, the performance of (1) and (2) is not sufficient when the playback volume is low, and (3) is satisfied when the playback volume is high. Can not.

  In both Non-Patent Document 1 and Non-Patent Document 2, it is assumed that the input acoustic signal is a signal in which a plurality of sounds such as music are mixed. However, when the input acoustic signal is a single voice such as an announcement, for example, sound quality deterioration due to embedding a digital watermark becomes remarkable, and it becomes more difficult to satisfy the condition (3).

  As a similar system example, the configuration shown in FIG. 2 can be considered. The difference from the configuration of FIG. 1 is that the input sound signal is separately sent as “original sound” to the digital watermark reading unit 96 by the digital communication means 97. Satisfying the above (1), (2), and (3) with the configuration of FIG. 2 is easier than the configuration of FIG. The reason is that since the “original sound” is known, it is easy to detect the difference between the collected sound signal and the “original sound (input sound signal)”. The configuration in FIG. 2 is used for copyright management (embedding copyright information and detecting unauthorized copying), etc., but for use in the applications and services described in the background art, a transmission unit 91 and a reception unit A system capable of digitally communicating with 94 is required. If there is a mechanism capable of digital communication, parallel information may be sent via a digital communication path, and the configuration of FIG. 2 cannot achieve the object of the present invention.

  SUMMARY OF THE INVENTION An object of the present invention is to provide an acoustic digital watermarking system that realizes ensuring the communication speed of parallel information, ensuring communication reliability, and maintaining sound quality without necessarily going through a digital communication path.

In order to solve the above problems, according to one aspect of the present invention, an acoustic digital watermark system includes a digital watermark embedding unit and a digital watermark reading unit. The digital watermark embedding unit uses a FIR filter that uses a first linear prediction coefficient that is a linear prediction coefficient of the input acoustic signal as a filter coefficient to filter the input acoustic signal and obtain a first residual signal; ,
A first carrier generation unit that generates a sine wave that is N times the first fundamental frequency, which is the fundamental frequency of the first residual signal, and N is one or more integers, and M is one or more. A phase modulation unit that obtains a modulated wave by phase-modulating the first carrier with parallel information so that the symbol rate is M times the pitch frequency, and a modulated wave and a first residual signal. And an adder for obtaining the first signal, and a synthesis filter for obtaining the reproduced signal by filtering the first signal using an IIR filter having the first linear prediction coefficient as a filter coefficient. The digital watermark reading unit filters the collected sound signal using an FIR filter that uses a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal obtained by collecting the reproduced sound of the reproduced signal as a filter coefficient, A second inverse filter for obtaining a second residual signal, a second carrier generation unit that generates a sine wave that is N times the second fundamental frequency, which is the fundamental frequency of the second residual signal, as a second carrier, And a detector that detects information corresponding to the parallel information using the residual signal and the second carrier wave.

  In order to solve the above-described problem, according to another aspect of the present invention, an electronic watermark embedding apparatus uses an FIR filter that uses a first linear prediction coefficient that is a linear prediction coefficient of an input acoustic signal as a filter coefficient, A first inverse filter that filters the input acoustic signal to obtain a first residual signal, and N is an integer greater than or equal to 2 and is N times the first fundamental frequency, which is the fundamental frequency of the first residual signal. A first carrier wave generation unit that generates a sine wave as the first carrier wave, and M is any integer between 1 and N, and the first carrier wave is phased in parallel information so that the symbol rate is M times the pitch frequency. A first phase modulation unit that performs modulation and obtains a modulated wave, an addition unit that obtains a first signal by adding the modulated wave and the first residual signal, and an IIR filter that uses the first linear prediction coefficient as a filter coefficient Filter the first signal Ngushi, and a synthesis filter for obtaining a reproduced signal.

  In order to solve the above-described problem, according to another aspect of the present invention, a digital watermark reading apparatus obtains parallel information using a collected sound signal. The digital watermark reading apparatus uses a FIR filter that uses a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal as a filter coefficient to filter the collected sound signal and obtain a second residual signal; , N is any integer of 2 or more, and a second carrier generation unit that generates a sine wave that is N times the second fundamental frequency, which is the fundamental frequency of the second residual signal, as a second carrier, and a second residual And a detection unit that detects information corresponding to the parallel information using the difference signal and the second carrier wave.

  In order to solve the above problem, according to another aspect of the present invention, an acoustic watermarking method uses an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of an input acoustic signal as a filter coefficient, A first inverse filtering step of filtering an input acoustic signal to obtain a first residual signal, N being an integer of 2 or more, and N times a first fundamental frequency that is a fundamental frequency of the first residual signal; A first carrier wave generating step for generating a sine wave as a first carrier wave, and M is any integer from 1 to N, and the first carrier wave is set in parallel information so that the symbol rate is M times the pitch frequency. A phase modulation step for performing phase modulation and obtaining a modulated wave, an addition step for obtaining the first signal by adding the modulated wave and the first residual signal, and an IIR filter using the first linear prediction coefficient as a filter coefficient are used. A first filtering step for filtering the first signal to obtain a reproduction signal, and an FIR using a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal obtained by collecting the reproduced sound of the reproduced signal as a filter coefficient A second inverse filtering step of filtering the collected sound signal to obtain a second residual signal using a filter; and a second sine wave that is N times the second fundamental frequency, which is the fundamental frequency of the second residual signal. A second carrier generation step for generating the carrier wave; and a detection step for detecting information corresponding to the parallel information using the second residual signal and the second carrier wave.

  In order to solve the above problem, according to another aspect of the present invention, a digital watermark embedding method uses an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of an input acoustic signal as a filter coefficient, A first inverse filtering step of filtering an input acoustic signal to obtain a first residual signal, N being an integer of 2 or more, and N times a first fundamental frequency that is a fundamental frequency of the first residual signal; A first carrier wave generating step for generating a sine wave as a first carrier wave, and M is any integer from 1 to N, and the first carrier wave is set in parallel information so that the symbol rate is M times the pitch frequency. Phase modulation, a first phase modulation step for obtaining a modulated wave, an addition step for adding the modulated wave and the first residual signal to obtain a first signal, and an IIR filter using the first linear prediction coefficient as a filter coefficient Using data, filtering the first signal, and a synthesis filtering step of obtaining a reproduced signal.

  In order to solve the above-described problem, according to another aspect of the present invention, a digital watermark reading method obtains parallel information using a collected sound signal. In the digital watermark reading method, a second inverse filtering step of obtaining a second residual signal by filtering the collected sound signal using an FIR filter having a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal as a filter coefficient A second carrier generation step for generating a sine wave that is N times the second fundamental frequency, which is the fundamental frequency of the second residual signal, as a second carrier, where N is any integer greater than or equal to 2; A detection step of detecting information corresponding to the parallel information using the residual signal and the second carrier wave.

  According to the present invention, there is an effect that it is possible to achieve the communication speed of parallel information, the reliability of communication, and the maintenance of voice quality without necessarily going through a digital communication path.

The figure which shows the structural example of an acoustic digital watermark system. The figure which shows the structural example of an acoustic digital watermark system. The functional block diagram of the electronic watermark embedding part which concerns on 1st embodiment. The figure which shows the example of the processing flow of the electronic watermark embedding part which concerns on 1st embodiment. The functional block diagram of the electronic watermark reading part which concerns on 1st embodiment. The figure which shows the example of the processing flow of the electronic watermark reading part which concerns on 1st embodiment. FIG. 7A is a diagram showing a signal obtained by modeling a residual signal with a simple pulse train, and FIG. 7B is a diagram showing a signal obtained by passing the signal of FIG. 7A through a synthesis filter. The figure which shows the frequency spectrum which Fourier-transformed the signal of FIG.7 (B). 9A shows a signal obtained by adding a carrier wave to FIG. 7A, and FIG. 9B shows a signal obtained by passing the signal of FIG. 9A through a synthesis filter. The figure which shows the frequency spectrum which carried out the Fourier-transform of the signal of FIG.9 (B). 11A shows a signal obtained by adding a carrier wave having a phase different from that of FIG. 9A to FIG. 7A, and FIG. 11B shows a signal obtained by passing the signal of FIG. 11A through a synthesis filter. FIG. The figure which shows the frequency spectrum which Fourier-transformed the signal of FIG. 13A shows a signal obtained by modulating the carrier wave of FIG. 9A with BPSK, and FIG. 13B shows a signal obtained by passing the signal of FIG. 13A through a synthesis filter. The figure which shows the frequency spectrum which carried out the Fourier-transform of the signal of FIG.13 (B). FIG. 15A shows a signal obtained by setting the carrier frequency and symbol rate to a value that is not an integral multiple of the pitch frequency, and FIG. 15B shows a signal obtained by adding the signal of FIG. 15A to the synthesis filter. FIG. The figure which shows the frequency spectrum which Fourier-transformed the signal of FIG.15 (B). The functional block diagram of the electronic watermark embedding part which concerns on 2nd embodiment. The figure which shows the example of the processing flow of the electronic watermark embedding part which concerns on 2nd embodiment.

Hereinafter, embodiments of the present invention will be described. In the drawings used for the following description, constituent parts having the same function and steps for performing the same process are denoted by the same reference numerals, and redundant description is omitted.
<Acoustic Watermarking System According to First Embodiment>
The acoustic digital watermark system 10 according to the first embodiment includes a transmission unit 11 and a reception unit 14. The transmission unit 11 and the reception unit 14 each include a digital watermark embedding unit 100 and a digital watermark reading unit 200 (see FIG. 1).

  The digital watermark embedding unit 100 of the transmission unit 11 receives the input acoustic signal and the parallel information, and outputs a reproduction signal. A reproduction signal is reproduced by the speaker 93 of the transmission unit 11 and a reproduction sound is emitted. The speaker 93 and the microphone 95 are arranged in a common sound field.

  The microphone 95 of the receiving unit 14 collects the reproduced sound and outputs a collected sound signal. The digital watermark reading unit 200 of the receiving unit 14 receives the collected sound signal and outputs an output acoustic signal and information corresponding to the parallel information. For example, the parallel information may be output and displayed on a display unit such as a display or a touch panel. When the listener is in a place where the reproduced sound of the speaker 93 can be sufficiently heard, the digital watermark reading unit 200 may output only parallel information without outputting the output acoustic signal.

Details of the digital watermark embedding unit 100 and the digital watermark reading unit 200 will be described below.
<Digital Watermark Embedding Unit According to First Embodiment>
FIG. 3 is a functional block diagram of the digital watermark embedding unit 100 according to the first embodiment, and FIG. 4 shows its processing flow.

  The digital watermark embedding unit 100 includes an input buffer 110, an inverse filter 130, a linear prediction analysis unit 120, a pitch analysis unit 140, a carrier wave generation unit 150, a phase modulation unit 160, an addition unit 170, and a synthesis filter 180.

<Input buffer 110>
The input buffer 110 receives the input sound signal, accumulates the input sound signal for at least a certain time (S110), and outputs the input sound signal every certain time (frame). In other words, the input acoustic signal is stored in the input buffer 110, and the input acoustic signal is sent to the linear prediction analysis unit 120 and the inverse filter 130 after being divided at regular time intervals called frames. In general, the time length of one frame is generally about 10 to 20 milliseconds, but other time lengths may be used. Hereinafter, the input acoustic signal divided at regular intervals is also referred to as a frame signal.

<Linear prediction analysis unit 120>
The linear prediction analysis unit 120 receives the frame signal, performs linear prediction analysis on the input acoustic signal using a linear prediction analysis technique, obtains a linear prediction coefficient (S120), and converts the linear prediction coefficient into the inverse filter 130 and the synthesis filter 180. Send to. As a method of linear prediction analysis, a covariance method, an autocorrelation method, a PARCOR analysis, an LSP analysis, or the like can be considered. Note that the linear prediction coefficient includes a value equivalent to the linear prediction coefficient (for example, PARCOR coefficient, LSP (line spectrum pair)) in addition to the linear prediction coefficient itself.

<Inverse filter 130>
The inverse filter 130 receives the frame signal and the linear prediction coefficient, filters the input acoustic signal using an FIR filter using the linear prediction coefficient as a filter coefficient (S130), obtains a residual signal, and outputs it. The residual signal is sent to the pitch analysis unit 140 and the addition unit 170.

<Pitch analysis unit 140>
The pitch analysis unit 140 receives the residual signal, analyzes the pitch length of the residual signal, that is, the length of one period of the fundamental frequency of the speech (S140), and sends the pitch frequency or the pitch length to the carrier wave generation unit 150. . The pitch frequency and the pitch length have a reciprocal relationship, and each has a unit of Hz and second. Unless otherwise specified, the pitch frequency and the pitch length are synonymous, and represent the fundamental frequency in the frequency domain and the time length of the pitch in the time domain. Hereinafter, they are collectively referred to as “pitch”.

<Carrier Generation Unit 150>
The carrier wave generation unit 150 receives the pitch, generates a sine wave whose frequency is N times the pitch frequency (one cycle is 1 / N of the pitch length) as a carrier wave (S150), and outputs it. N is any integer of 2 or more. For example, when the pitch is 100 Hz, a sine wave of 500 Hz, 800 Hz, 1000 Hz or the like is output. The upper limit of N is not particularly limited, but the effect of the present embodiment is great if N times the pitch frequency is determined to be 4 kHz or less. It is known that an acoustic signal such as a voice or a musical instrument sound has a “harmonic structure” composed of a pitch frequency component and its harmonic component (also referred to as a harmonic component). Since the harmonic structure is remarkable in the following frequency components and the ratio of the frequency components of 4 kHz or less is larger than the ratio of the frequency components of 4 kHz or more, the auditory auditory sense of the listener by superimposing the carrier wave is increased. This is because distortion can be suppressed. Note that the phase of the carrier wave matches the phase of the pitch frequency component of the residual, or the relative phase difference with the phase of the pitch frequency component of the residual is always a constant value.

<Phase modulation unit 160>
The phase modulation unit 160 receives the carrier wave and the parallel information, phase-modulates the carrier wave with the parallel information so that the symbol rate is M times the pitch frequency, obtains a modulated wave (S160), and outputs it. Note that M is any integer from 1 to N. The parallel information is a digital signal represented by a time series of 1 and 0. BPSK (Binary Phase Shift Keying) and QPSK (Quadrature Phase Shift Keying) are known as phase modulation methods. BPSK is also called 2PSK and QPSK is also called 4PSK. In BPSK, 1 symbol can transmit 1 bit, and in QPSK, 1 symbol can transmit 2 bits of parallel information.

<Adding unit 170>
The adder 170 receives the modulated wave and the residual signal, adds the modulated wave and the residual signal (S170), and adds the added signal (hereinafter also referred to as “first signal”) to the synthesis filter 180. send.

<Synthesis filter 180>
The synthesis filter 180 receives the linear prediction coefficient and the first signal, filters the first signal using an IIR filter using the linear prediction coefficient as a filter coefficient (S180), obtains and outputs a reproduction signal.

Details regarding the linear prediction analysis, the inverse filter, the synthesis filter 180, and the pitch analysis are described in Reference Document 1, for example. Details of the carrier wave and the phase modulation are described in Reference Document 2, for example.
[Reference 1] Sadaaki Furui, “Digital Audio Processing”, Tokai University Press, 1985, pp.60-98
[Reference 2] Yukihiro Kamiya, "Digital wireless communication technology using C language", Corona, 2010, pp.53-83

<Digital Watermark Reading Unit According to First Embodiment>
FIG. 5 is a functional block diagram of the digital watermark reading unit 200 according to the first embodiment, and FIG. 2 shows a processing flow thereof.

  The digital watermark reading unit 200 includes an input buffer 210, an inverse filter 230, a linear prediction analysis unit 220, a pitch analysis unit 240, a carrier wave generation unit 250, and a detection unit 260.

  The collected sound signal is sent to the input buffer 210 and output as an output sound signal without processing.

  The input buffer 210, the linear prediction analysis unit 220, the inverse filter 230, the pitch analysis unit 240, and the carrier wave generation unit 250 are the linear prediction analysis unit 120, the inverse filter 130, the pitch analysis unit 140, and the carrier wave generation unit of the digital watermark embedding unit 100 of FIG. The portion that corresponds to the operation of the unit 150 and corresponds to the addition unit 170 of FIG. The outline will be described below.

<Input buffer 210>
The input buffer 210 receives the collected sound signal, accumulates the collected sound signal for at least a certain time (S210), and outputs the collected sound signal for every certain time (frame).

<Linear prediction analysis unit 220>
The linear prediction analysis unit 220 receives the collected sound signal for each frame, performs linear prediction analysis of the collected sound signal using a linear prediction analysis technique, obtains a linear prediction coefficient (S220), and uses the linear prediction coefficient as an inverse filter 230. Send to.

<Inverse filter 230>
The inverse filter 230 receives the collected sound signal and the linear prediction coefficient for each frame, filters the collected sound signal using an FIR filter using the linear prediction coefficient as a filter coefficient (S230), obtains a residual signal, and outputs it. To do. The residual signal is sent to the pitch analysis unit 240 and the detection unit 260.

<Pitch analysis unit 240>
The pitch analysis unit 240 receives the residual signal, analyzes the pitch length of the residual signal, that is, the length of one period of the fundamental frequency of the voice (S240), and sends the pitch to the carrier wave generation unit 250.

<Carrier Generation Unit 250>
The carrier wave generation unit 250 receives the pitch, generates a sine wave whose frequency is N times the pitch frequency as a carrier wave (S250), and outputs it.

<Detection unit 260>
The detector 260 receives the residual signal and the carrier wave, detects information corresponding to the parallel information using the residual signal and the carrier wave in which the digital watermark is embedded, and outputs it (S260). If the communication is normally performed, the parallel information embedded by the digital watermark embedding unit 100 matches the information corresponding to the parallel information detected in S260. As a detection method, a common phase modulation type synchronous detection method can be used. Here, the point is that the method of synchronous detection can be used. In general, an advanced digital communication network can generate a carrier wave synchronized between a transmission side and a reception side, so that synchronous detection is possible, and high-speed and stable digital communication is realized. However, in a simple digital communication system, delay detection or asynchronous detection is mainstream, and inferior to synchronous detection in terms of communication stability. In the present embodiment, synchronous detection can be performed in spite of a simple modulation / demodulation method. The carrier wave is synchronized with both the digital watermark embedding unit 100 and the digital watermark reading unit 200 in frequency and phase with the pitch of the residual signal. This is because, as a result, a carrier wave synchronized between the transmission side and the reception side can be generated.

Details regarding synchronous detection, delay detection, and asynchronous detection are described in Reference Document 3, for example.
[Reference 3] Yoichi Saito, "Modulation and Demodulation of Digital Wireless Communication", IEICE, 1996, pp.1-35

<Effect>
With the above configuration, by analyzing the pitch frequency of the input acoustic signal and superimposing it on the input acoustic signal as a watermark with a phase modulation signal with a frequency that is an integral multiple of the pitch frequency as a carrier wave and a symbol rate that is an integral multiple of the pitch frequency. An acoustic digital watermarking system that realizes (1) ensuring the communication speed of parallel information, (2) ensuring communication reliability, and (3) maintaining voice quality without necessarily going through a digital communication path.

  It will be described that the present embodiment satisfies the above (1), (2), and (3).

  As for the bit rate of the parallel information that can be transmitted in (1), if the pitch frequency is 100 Hz and the symbol rate is 200 Hz, parallel information of 200 bps (bits per second) can be transmitted when BPSK is applied, and 400 bps can be transmitted when QPSK is applied. If the content of the utterance is converted to text at the speed at which a person speaks, it is at most about 100 bps, and a sufficient speed can be obtained.

  As described above, (2) can perform synchronous detection based on the pitch, and can read parallel information stably.

  What is important is (3). When playing back a playback signal or output acoustic signal, it is completely meaningless if the modulated wave added to the residual signal is heard as an annoying sound. This will be shown using a specific example.

  It is known that a vowel interval of speech is linearly predicted, and the obtained residual signal has a pulse-like waveform. For simplicity, description will be made by modeling with a simple pulse train whose interval is the pitch length.

  FIG. 7 (A) assumes that the residual signal in the vowel section of speech is a simple pulse train with a pitch frequency of 125 Hz. The pulses are arranged at equal intervals of 8 milliseconds corresponding to 125 Hz.

  FIG. 7B shows a signal obtained by passing the signal shown in FIG. 7A through a synthesis filter having a linear prediction coefficient obtained as a filter coefficient obtained by analyzing the speech uttered as “A”. Since the residual signal is approximated by a simple pulse, it is somewhat different from the original “A” acoustic signal, but when the signal in FIG. 7B is reproduced by a speaker, a sound “A” is heard.

  FIG. 8 shows a frequency spectrum (power spectrum) obtained by performing Fourier transform (FFT) on the signal shown in FIG. The spectrum has a form in which harmonics of an integral multiple of 125 Hz are modulated with a spectrum envelope which is a frequency characteristic of the synthesis filter.

  FIG. 9A shows a signal obtained by adding a carrier wave to FIG. The frequency of the carrier wave was set to 8 times the pitch frequency, that is, 1 kHz.

  FIG. 9B is a signal obtained by passing FIG. 9A through the synthesis filter, and FIG. 10 is a frequency spectrum of FIG. 9B.

  Comparing FIG. 8 and FIG. 10, it can be seen that the amplitude of 1 kHz is increased by the amount of the carrier wave, but if you do not compare carefully, you will not notice the difference. Even if you actually listen to and compare FIGS. 7 (B) and 9 (B) by listening to the speakers, you will feel a slight difference in the timbre, but if you listen only to FIG. do not feel.

  11A is also a signal obtained by adding a carrier wave to FIG. 7A, as in FIG. 9A. The only difference between FIG. 9A and FIG. 11A is that the phase of the carrier wave is different by π / 2. Similarly, FIG. 11B is a signal that has passed through the synthesis filter, and FIG. 12 is a frequency spectrum. However, since the power spectrum is not related to the phase, FIGS. 10 and 12 are the same. However, there are some slight differences in terms of Fourier transform. Since the human ear is insensitive to the phase, the sounds in FIGS. 9B and 11B can be heard as the same sound.

  In this embodiment, even if the phase of the carrier wave is different, no difference is felt to the human ear. Therefore, as described above, the phase of the carrier wave is a relative phase difference from the phase of the pitch frequency component of the residual signal. However, if the digital watermark embedding unit 100 and the digital watermark reading unit 200 match, an arbitrary phase may be used.

  FIG. 13A shows a signal obtained by modulating the carrier wave shown in FIG. 9A with parallel information using the BPSK technique. As an example of the parallel information, 1 and 0 are alternated. The symbol rate was 250 Hz, which is twice the pitch frequency. FIG. 13B is a signal obtained by passing FIG. 13A through the synthesis filter, and FIG. 14 is a frequency spectrum thereof. Comparing FIG. 14 and FIG. 10, it can be seen that the difference is still slight. When listening to FIG. 13B by reproducing with a speaker, a slight difference in timbre is felt, but a sense of incongruity is not felt.

  Thus, even if a carrier wave having N times the pitch frequency (8 times in the above example) is modulated at a symbol rate M times the pitch frequency (2 times in the above example) and added to the residual signal, there is no sense of incongruity. You can get no playback sound. In the example of FIGS. 9A to 13A, the amplitude of the carrier wave is set to a large value (−14 dB of the amplitude of the residual signal) for the sake of easy understanding. However, when actually used, The amplitude of the carrier wave can be set sufficiently smaller than the amplitude of the residual signal, and the difference in timbre when reproduced by a speaker can be reduced.

  Next, a case where the frequency of the carrier wave or the symbol rate is not an integral multiple of the pitch frequency will be described.

  FIG. 15A shows a signal obtained by setting the carrier frequency or symbol rate to a value which is not an integer multiple of the pitch frequency, FIG. 15B shows a signal obtained by passing it through a synthesis filter, and FIG. 16 shows its frequency. It is a spectrum. Comparing FIG. 16 with FIG. 14, it can be seen that the harmonic structure is greatly collapsed and has a completely different spectrum. When the signal shown in FIG. 15B is reproduced by a speaker and listened to, the sound becomes very unpleasant and the same effect as in the present embodiment cannot be obtained.

<Second embodiment>
A description will be given centering on differences from the first embodiment.

<Acoustic Watermarking System According to Second Embodiment>
The acoustic digital watermark system 30 according to the second embodiment includes a transmission unit 31 and a reception unit 34. The transmission unit 31 and the reception unit 34 include a digital watermark embedding unit 300 and a digital watermark reading unit 400, respectively (see FIG. 1).

Details of the digital watermark embedding unit 300 will be described below.
<Digital Watermark Embedding Unit According to Second Embodiment>
FIG. 17 is a functional block diagram of the digital watermark embedding unit 300, and FIG. 18 shows its processing flow.

  The digital watermark embedding unit 300 includes an input buffer 110, an inverse filter 130, a linear prediction analysis unit 120, a pitch analysis unit 340, a carrier wave generation unit 150, a phase modulation unit 160, an addition unit 370, a synthesis filter 180, a determination unit 390, and a SW 391. including.

  All acoustic signals do not necessarily have a frequency characteristic (harmonic structure) like a vowel section of speech. For example, listening to the sound that has passed through the synthesis filter 180 by adding the carrier wave to the residual signal in a consonant section or a silent section, the carrier wave is heard, and a sense of incongruity is felt. Therefore, in this embodiment, the modulation wave and the residual signal are added only when the residual signal has a harmonic structure.

<Pitch analysis unit 340>
The pitch analysis unit 340 receives the residual signal, calculates the pitch frequency using the residual signal, outputs it to the carrier wave generation unit 150, and calculates the autocorrelation value of the residual signal at the pitch frequency (S340), It outputs to the determination part 390.

<Determining unit 390>
The determination unit 390 receives the autocorrelation value, uses this value to determine whether or not it has a frequency characteristic (harmonic structure) such as a vowel section, and controls the SW 391 on / off according to the determination result Output a control signal. When the frequency characteristic is the same as that of the vowel section, SW 391 is turned on and the carrier wave is added to the residual signal. If it is determined that the frequency characteristic is not the same as that of the vowel section, the SW 391 is turned OFF and the carrier wave is not added to the residual signal.

<Adding unit 370>
When the residual signal has a harmonic structure, the adding unit 370 adds the modulated wave and the residual signal to obtain a first signal (yes in S390, S370), and the residual signal does not have a harmonic structure. In this case, the residual signal is set as the first signal (no in S390). Adder 370 outputs the first signal to adder 170.

<Digital Watermark Reading Unit According to Second Embodiment>
The digital watermark reading unit 400 differs from the digital watermark reading unit 200 in the following points. The digital watermark reading unit 400 includes a detection unit 460 instead of the detection unit 260 (see FIG. 5).

  The detection unit 460 receives the residual signal and the carrier wave, and uses the residual signal and the carrier wave in which the digital watermark is embedded, and when the modulation wave is added on the transmission side, information corresponding to the parallel information is obtained. Detect (S460 in FIG. 6) and output. The digital watermark reading unit 400 needs to determine from the collected sound signal whether the SW 391 of the digital watermark embedding unit 300 is ON or OFF, that is, whether a modulated wave is added or not. There is. However, it is not necessary to perform the same processing as that of the determination unit 390. Since the carrier wave of the carrier wave generation unit 250 and the residual signal of the inverse filter 230 are uncorrelated, if the modulation wave is added as a result of detection by the detection unit 460, a baseband signal can be obtained with sufficient intensity, If not added, a baseband signal cannot be obtained with sufficient strength. Therefore, the digital watermark reading unit 400 determines that a modulated wave is added on the transmission side when the obtained baseband signal is equal to or higher than a predetermined intensity, and detects information corresponding to the parallel information. Output. On the other hand, when the obtained baseband signal is less than a predetermined strength, the digital watermark reading unit 400 determines that no modulated wave is added on the transmission side and does not output information corresponding to the parallel information. What should I do?

<Effect>
By setting it as such a structure, the effect similar to 1st embodiment can be acquired. Furthermore, a digital watermark is embedded only in a section where the reproduced sound does not feel uncomfortable even when the carrier wave is added.

  In addition, since the sound of stringed instruments and wind instruments has the same frequency characteristics (harmonic structure) as the voice vowel sections, the digital watermark can be embedded without any sense of incongruity with the sounds of stringed instruments or wind instruments. be able to.

<Other variations>
The digital watermark embedding units 100 and 300 and the digital watermark reading units 200 and 400 may be configured as separate devices, and may be a digital watermark embedding device and a digital watermark reading device, respectively.

  In a communication system using digital modulation / demodulation, an error correction code and an error detection code are generally used together. In the present invention, the reliability of communication is further improved by using an error correction code and an error detection code together. Can be increased.

  The present invention is not limited to the above-described embodiments and modifications. For example, the various processes described above are not only executed in time series according to the description, but may also be executed in parallel or individually as required by the processing capability of the apparatus that executes the processes. In addition, it can change suitably in the range which does not deviate from the meaning of this invention.

<Program and recording medium>
The present invention can also be realized by being mounted on a digital signal processor or a dedicated LSI. Further, it can be executed as a computer main body and a computer program. In other words, various processing functions in each device described in the above embodiments and modifications may be realized by a computer. In that case, the processing contents of the functions that each device should have are described by a program. Then, by executing this program on a computer, various processing functions in each of the above devices are realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used.

  The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its storage unit. When executing the process, this computer reads the program stored in its own storage unit and executes the process according to the read program. As another embodiment of this program, a computer may read a program directly from a portable recording medium and execute processing according to the program. Further, each time a program is transferred from the server computer to the computer, processing according to the received program may be executed sequentially. Also, the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition. It is good. Note that the program includes information provided for processing by the electronic computer and equivalent to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

  In addition, although each device is configured by executing a predetermined program on a computer, at least a part of these processing contents may be realized by hardware.

Claims (8)

Including a digital watermark embedding unit and a digital watermark reading unit,
The digital watermark embedding unit includes:
A first inverse filter that obtains a first residual signal by filtering the input acoustic signal using an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of the input acoustic signal as a filter coefficient;
A first carrier generation unit that generates N as a first carrier wave a sine wave that is N times the first fundamental frequency, which is a fundamental frequency of the first residual signal, and N is any integer of 2 or more;
A phase modulation unit that obtains a modulated wave by phase-modulating the first carrier wave with parallel information so that M is any integer from 1 to N and the symbol rate is M times the pitch frequency;
An adder for adding the modulated wave and the first residual signal to obtain a first signal;
Using an IIR filter having the first linear prediction coefficient as a filter coefficient, and filtering the first signal to obtain a reproduction signal;
The digital watermark reading unit
Filtering the collected sound signal using an FIR filter using a second linear prediction coefficient, which is a linear prediction coefficient of the collected sound signal obtained by collecting the reproduced sound of the reproduced signal, as a filter coefficient; A second inverse filter for obtaining a signal;
A second carrier generation unit that generates, as a second carrier wave, a sine wave that is N times the second fundamental frequency that is the fundamental frequency of the second residual signal;
A detector that detects information corresponding to the parallel information using the second residual signal and the second carrier wave;
Acoustic watermarking system. The acoustic watermarking system of claim 1,
When the first residual signal has a harmonic structure, the adding unit adds the modulated wave and the first residual signal to obtain the first signal, and the first residual signal is a harmonic structure. If the first residual signal is not the first signal,
Acoustic watermarking system. A first inverse filter that obtains a first residual signal by filtering the input acoustic signal using an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of the input acoustic signal as a filter coefficient;
A first carrier generation unit that generates N as a first carrier wave a sine wave that is N times the first fundamental frequency, which is a fundamental frequency of the first residual signal, and N is any integer of 2 or more;
A first phase modulation unit that obtains a modulated wave by phase-modulating the first carrier wave with parallel information so that M is any integer from 1 to N and the symbol rate is M times the pitch frequency;
An adder for adding the modulated wave and the first residual signal to obtain a first signal;
Using an IIR filter having the first linear prediction coefficient as a filter coefficient, and filtering the first signal to obtain a reproduction signal;
Digital watermark embedding device. A digital watermark reading device that obtains parallel information using a collected sound signal,
A second inverse filter for filtering the collected sound signal to obtain a second residual signal using an FIR filter having a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal as a filter coefficient;
A second carrier wave generation unit that generates N as a second carrier wave a sine wave that is N times a second fundamental frequency that is a fundamental frequency of the second residual signal, and N is any integer of 2 or more,
A detector for detecting information corresponding to parallel information using the second residual signal and the second carrier wave;
Digital watermark reader. A first inverse filtering step of filtering the input acoustic signal to obtain a first residual signal using an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of the input acoustic signal as a filter coefficient;
A first carrier generation step for generating a sine wave that is N times the first fundamental frequency, which is the fundamental frequency of the first residual signal, as a first carrier, where N is any integer greater than or equal to 2;
A phase modulation step for obtaining a modulated wave by phase-modulating the first carrier wave with parallel information so that M is any integer from 1 to N and the symbol rate is M times the pitch frequency;
Adding the modulated wave and the first residual signal to obtain a first signal;
Using an IIR filter having the first linear prediction coefficient as a filter coefficient, a synthetic filtering step of filtering the first signal to obtain a reproduction signal;
Filtering the collected sound signal using an FIR filter using a second linear prediction coefficient, which is a linear prediction coefficient of the collected sound signal obtained by collecting the reproduced sound of the reproduced signal, as a filter coefficient; A second inverse filtering step for obtaining a signal;
A second carrier generation step of generating, as a second carrier wave, a sine wave that is N times the second fundamental frequency that is the fundamental frequency of the second residual signal;
Using the second residual signal and the second carrier wave to detect information corresponding to the parallel information,
Acoustic watermarking method. A first inverse filtering step of filtering the input acoustic signal to obtain a first residual signal using an FIR filter having a first linear prediction coefficient that is a linear prediction coefficient of the input acoustic signal as a filter coefficient;
A first carrier generation step for generating a sine wave that is N times the first fundamental frequency, which is the fundamental frequency of the first residual signal, as a first carrier, where N is any integer greater than or equal to 2;
A first phase modulation step for obtaining a modulated wave by phase-modulating the first carrier wave with parallel information so that M is any integer from 1 to N and the symbol rate is M times the pitch frequency;
Adding the modulated wave and the first residual signal to obtain a first signal;
Using an IIR filter having the first linear prediction coefficient as a filter coefficient, and filtering the first signal to obtain a reproduction signal;
Digital watermark embedding method. A digital watermark reading method for obtaining parallel information using a collected sound signal,
A second inverse filtering step of filtering the collected sound signal to obtain a second residual signal using an FIR filter having a second linear prediction coefficient that is a linear prediction coefficient of the collected sound signal as a filter coefficient;
A second carrier generation step of generating N as a second carrier a sine wave that is N times the second fundamental frequency, which is a fundamental frequency of the second residual signal, and N is any integer of 2 or more;
Using the second residual signal and the second carrier wave to detect information corresponding to parallel information,
Digital watermark reading method.   A program for causing a computer to function as the digital watermark embedding device according to claim 3 or the digital watermark reading device according to claim 4.