JP5046233B2 - Speech enhancement processor - Google Patents

Description

  The present invention relates to a speech enhancement processing device for compressing a high frequency component of input speech to a low frequency range and providing clear and natural speech in a public broadcasting device, a loudspeaker, a hearing aid, a telephone, and the like.

  It is known that human hearing gradually decreases from hearing in a high frequency range with aging. This is a deafness caused by a decrease in the function of the inner ear, and is generally referred to as sound-sensitive deafness. Among them, those caused by aging are particularly referred to as senile deafness. In senile deafness, it has been statistically proven that the frequency at which hearing begins to decline with age is decreasing.

  In many cases, a decrease in hearing ability in a high frequency band is also observed in sensorineural hearing loss that is not senile deafness. It is no exaggeration to say that the majority of people who suffer from hearing problems, such as the elderly and hearing-impaired, suffer from hearing loss in the high frequency band.

  On the other hand, since the consonant part of human speech contains many high-frequency components, if the hearing ability in the high frequency range decreases, the ability to hear and discriminate the consonant part decreases, making it difficult to accurately understand the content of the speech. As a result, it causes a great obstacle to voice communication.

  Many of the hearing aids currently on the market have the function to amplify only the high frequency components of the input sound (high frequency emphasis). In many cases, it shows a decrease in frequency resolution and recruitment phenomenon (a phenomenon in which a low level sound cannot be heard but a high level sound feels noisy like ordinary people). In many cases, it cannot be obtained.

  Many digital hearing aids in recent years are equipped with a function called nonlinear amplification to compensate for the recruitment phenomenon. This is a function that amplifies low-level input sounds and does not amplify high-level sounds, and is a function that is supported by many elderly people and hearing-impaired people from the viewpoint of the comfort of hearing aid sounds.

  Patent Document 1 discloses a means for converting speech into a characteristic parameter related to frequency spectrum envelope, sound source frequency and sound source amplitude for a hearing impaired person whose hearing in the high frequency range has decreased, There is a description of a hearing aid characterized in that it comprises means for synthesizing and synthesizes and outputs a sound whose frequency spectrum envelope is compressed to a low frequency band.

  In addition, even if hearing in the high frequency range is reduced, there are many elderly people and hearing-impaired people who do not use such hearing aids due to their appearance and economic problems. It is not difficult for an elderly person or a hearing-impaired person who does not use these hearing aids to hear a sound such as an announcement in an environment where there is a lot of noise such as in a station.

On the other hand, in a telephone, the transmission frequency band is limited to 300 to 3400 Hz in order to reduce the amount of communication data. This means that in a telephone call, a general user also listens to a sound in which information in the high frequency range is cut, that is, the general user also has a hearing impaired person whose hearing in the high frequency range has decreased. It means that the user is listening to the sound in the same state. Actually, in the current commercially available telephones, there are not a few cases where it is difficult to accurately understand the contents of voice even if the elderly or the deaf.
JP-A-57-178499

  For the elderly and hearing-impaired people whose hearing in the high frequency range has been reduced, simply amplifying the sound level will cause the low frequency range component with normal hearing to be excessively heard and feel noisy. Further, it is known that even if high-frequency emphasis or nonlinear amplification is performed, sufficient hearing aid effects are often not obtained due to other factors such as frequency resolution.

  Patent Document 1 describes a frequency compression method for a hearing-impaired person whose hearing in the high frequency range has decreased. However, this method also compresses the harmonic structure of the sound, so that the naturalness of the sound is impaired. Therefore, there was a problem that it was difficult to use for elderly people and those with mild hearing loss.

  Also, in an environment where a lot of noise is mixed, such as in a station premises, high frequency emphasis or non-linear amplification may be applied to the announcement speech for elderly people who are not using hearing aids or for hearing-impaired people. A person's auditory characteristics vary from person to person, and it has been difficult to uniformly apply such an audio signal processing method for a hearing aid.

  Furthermore, in the case of telephones, it is practically difficult to expand the transmission frequency band from the viewpoint of communication costs and infrastructure development, but in the current narrow transmission frequency band, it is highly clear and natural. It was difficult to realize a high quality voice call.

  In order to solve the above-described problems, the present invention has the following configuration regarding a speech enhancement processing device for providing clear and natural speech in a public broadcasting device, a loudspeaker, a hearing aid, a telephone, and the like.

  An input band limiting unit, an input band power calculating unit, a harmonic code calculating unit, an in-band signal multiplying unit, an output signal limiting unit, and a signal adding unit, wherein the input band limiting unit has a predetermined frequency of the input signal. A first band-pass filter that passes the component as it is, and an input in-band power calculation unit that sets a lower limit frequency to a frequency value equal to or higher than an upper limit frequency of the first band filter and passes the input signal. And a power envelope extraction unit that extracts a power envelope signal of an output signal of the second band filter, and the harmonic code calculation unit has an arbitrary pass frequency band, and the input signal And a code extraction unit that extracts a code of the output signal of the harmonic extraction filter, and the in-band signal multiplication unit A compression band filter having a function of multiplying the output signal of the word calculation unit and the harmonic code calculation unit, wherein the output signal limiting unit uses a frequency band in which a frequency component of the input signal is compressed as a pass frequency band The signal adding unit has a function of adding the output of the input band limiting unit and the output of the output signal limiting unit. Thereby, after compressing high frequency components, such as a consonant of an input audio | voice, to a low frequency range, clear and natural audio | voice can be provided.

  Further, the pass frequency band of the harmonic extraction filter coincides with the first band filter. As a result, the harmonicity of the signal in the frequency-compressed band can be maintained and clear and natural sound can be provided, and the first bandpass filter and the harmonic extraction filter can be the same filter. It can be configured with a smaller system.

  Further, the pass frequency band of the harmonic extraction filter coincides with the compression band filter. As a result, the harmonics of the signal in the frequency-compressed band can be more accurately maintained, and clear and natural sound can be provided. In addition, the first band filter and the harmonic extraction filter can be provided. It can be configured with the same filter, and can be configured with a smaller system.

  The input in-band power calculating unit, the harmonic code calculating unit, the in-band signal multiplying unit, and the output signal limiting unit each have a plurality, and each input in-band power calculating unit passes through the second filter. The band is not overlapped, and the pass band of the compression band filter in each output signal limiting unit is not overlapped. Thereby, clear voice can be provided without impairing the harmonic nature and naturalness of high frequency components such as consonants of voice.

  The output signal of the input in-band power calculation unit is amplified. As a result, it is possible to provide a clearer voice by compressing and emphasizing high frequency components such as consonants of the input voice for the elderly and hearing impaired persons whose hearing ability in the high frequency range has decreased.

  Further, the output signal of the output signal limiting unit is amplified. As a result, it is possible to provide a clearer voice by compressing and emphasizing high frequency components such as consonants of the input voice for the elderly and hearing impaired persons whose hearing ability in the high frequency range has decreased.

The input band limiting unit includes an input band power calculating unit, a mapping destination band power calculating unit, a power ratio calculating unit, a power ratio multiplying unit, and a signal adding unit. The first in-band power calculation unit has a lower limit frequency set to a frequency value equal to or higher than the upper limit frequency of the first band filter and passes the input signal. And a power envelope extraction unit that extracts a power envelope signal of the output signal of the second band filter, and the mapping target in-band power calculation unit compresses the frequency component of the input signal. The input signal compression band filter that passes the input signal, and the output signal of the input signal compression band filter A power envelope extraction unit that extracts a warp envelope signal, and the power ratio calculation unit has a function of calculating a ratio of output signals of the input band power calculation unit and the mapping destination band power calculation unit, The power ratio multiplier has a function of multiplying the input signal compression band filter and the output signal of the power ratio calculator, and the signal adder is an output of the input band limiter and an output of the power ratio multiplier It has the structure characterized by having the function to add. As a result, a high-frequency component such as a consonant of the input voice can be compressed to a low frequency range, and a clear and natural voice can be provided. Further, since the arithmetic processing is simple, a small scale and low power consumption can be achieved. The system can be configured.

  The input in-band power calculation unit, the mapping destination in-band power calculation unit, the power ratio calculation unit, and the power ratio multiplication unit each have a plurality, and each input in-band power calculation unit passes through the second filter. The band is not overlapped, and the pass band of the input signal compression band filter in each mapping destination band power calculation section is not overlapped. As a result, a clear voice can be provided without impairing the harmonics and naturalness of high-frequency components such as voice consonants, and a small-scale, low-power-consumption system because the arithmetic processing is simple. Can be configured.

Further, the output signal of the input signal compression band filter is amplified. As a result, it is possible to provide a clearer voice by compressing and emphasizing high frequency components such as consonants of the input voice for elderly people and hearing impaired persons whose hearing ability in the high frequency range has decreased. Since the arithmetic processing is simple, it can be configured with a small-scale and low power consumption system.

In addition, the output signal of the power ratio multiplication unit is amplified. As a result, it is possible to provide a clearer voice by compressing and emphasizing high frequency components such as consonants of the input voice for elderly people and hearing impaired persons whose hearing ability in the high frequency range has decreased. Since the arithmetic processing is simple, it can be configured with a small-scale and low power consumption system.

    The speech enhancement processing apparatus of the present invention compresses high frequency components of speech such as consonants toward a low frequency range, and excessively amplifies the level of speech to elderly people and hearing impaired people whose hearing in the high frequency range has been reduced. Therefore, it is possible to provide clear and natural speech only by information in the frequency range remaining in the auditory sense without enhancing the high frequency range of the speech. Furthermore, the speech enhancement processing device of the present invention clearly and naturally provides high-frequency components of speech such as consonants that have been difficult to hear due to the bandwidth limitation even during telephone conversations where the transmission frequency bandwidth is limited. I can do it.

  Conventionally, in such audio processing, by compressing the frequency component, the harmonic structure of the audio is broken and the naturalness of the audio is often significantly impaired. However, the speech enhancement processing device according to the present invention can provide natural and clear speech while maintaining the harmonic structure of speech.

  In addition, the speech enhancement processing device of the present invention can generally amplify a consonant component that contains many high-frequency components and has a lower level than the vowel as needed in the process of the enhancement processing. As a result, even an elderly person or a hearing-impaired person whose hearing ability in the high frequency range has decreased, can hear natural and clear sound.

  Furthermore, the speech enhancement processing apparatus of the present invention can transmit and communicate a speech information such as a consonant that is compressed outside the transmission frequency band of the telephone within the transmission frequency band. This makes it possible to improve the telephone call quality of telephones not only for elderly people and persons with hearing impairment who have decreased hearing in the high frequency range, but also for general users.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings. In the following description, components having the same function are denoted by the same reference numerals, and repeated description thereof is omitted.

  FIG. 1 is a block diagram of a system according to the first embodiment of the present invention, in which an input band limiting unit 1 including a first band filter 7 that passes a predetermined frequency component of input speech as it is, A second band filter 8 that is set to a frequency value equal to or higher than the upper limit frequency of the first band filter 7 and passes the input signal, and a power envelope that extracts the power envelope signal of the output signal of the second band filter 8 The input band power calculation unit 2 including the extraction unit 9, the harmonic extraction filter 10 having an arbitrary pass frequency band and passing the input signal, and the sign of the output signal of the harmonic extraction filter 10 A harmonic code calculation unit 3 including a code extraction unit 11 to be extracted, an in-band that multiplies the output of the in-band power calculation unit 2 and the output of the harmonic code calculation unit 3 An output signal limiting unit 5 comprising a signal multiplying unit 4, a compression band filter 12 having a frequency band in which a frequency component of the input signal is compressed as a pass frequency band, an output of the input band limiting unit 1, and the output signal limiting unit 5 comprises a signal adder 6 for adding the outputs of 5.

    In the speech enhancement processing device of the present invention, the output signal of the second band filter 7 is compressed within the band of the compression band filter 12 which is narrower than the second band filter 7, and the compressed signal Is added to the output of the first band filter 7 by the signal adder 6 to generate a sound in which the sound component in the high frequency range is compressed to the low frequency range.

  For compression, the power envelope (power temporal change) of the output of the second bandpass filter 7 is extracted by the power envelope extractor 9, and the code of the output of the harmonic extraction filter 10 is calculated by the code extractor 11. is doing. Here, in this embodiment, this sign is set to 1 when the instantaneous amplitude value is positive, 0 when it is 0, and -1 when it is negative.

  The outputs of the power envelope extraction unit 9 and the code extraction unit 11 are multiplied by the in-band signal multiplication unit 4 and passed through the compression band filter 12. As a result, amplitude modulation is applied while maintaining the harmonic and non-harmonic characteristics of the input signal, so that frequency-compressed sound with high clarity and naturalness can be generated while maintaining the harmonic structure of the input sound. can do.

  FIG. 2 shows, as a system block diagram in the second embodiment of the present invention, the in-band power calculator 2, the harmonic code calculator 3, the in-band signal multiplier 4, and the output signal limiter. 5 in which the passbands of the second filter 8 in each input band power calculation unit 2 do not overlap and the passbands of the compression band filter 12 in each output signal limiter 5 do not overlap. Show.

  In the second embodiment, the high-frequency component of the input sound is divided into a plurality of bands, and the components in the wider band are compressed into the narrower band by compressing and adding the respective in-band signals. Is possible.

  The harmonic extraction filter 10 is a filter for encoding information related to the harmonic of the input signal by the code extraction unit 11. In this embodiment, the harmonic extraction filter 10 is configured as an independent filter having an arbitrary pass frequency band, but is shared with the first band filter 7 in order to simplify the system configuration. It is also possible. In this case, the output of the first band filter 7 is sent to the signal adder 6 and also sent to the code extraction unit 11 to extract the code, and each in-band signal multiplication unit 4 outputs each input in-band power calculation unit. Multiplied by the output of 2.

In FIG. 3, in the second embodiment, there are three types of the in-band power calculation unit 2, the harmonic code calculation unit 3, the in-band signal multiplication unit 4, and the output signal limiting unit 5, respectively. Furthermore, a signal aspect in the case where the pass frequency bands of the harmonic extraction filter 10 and the compression band filter 12 match is schematically shown. Here, the numbers shown in bold in the figure are the symbols of the corresponding components.
Hereinafter, the present embodiment will be described more specifically with reference to FIG.

  In this embodiment, the boundary frequency (lower limit frequency−upper limit frequency) of the first bandpass filter 7 is 70-3150 Hz, and the boundary frequency of the second bandpass filter 8 is (1) 3150-4000 Hz, (2) 4000- The boundary frequency of the harmonic extraction filter 10 is (1) 3150-3400 Hz, (2) 3400-3650 Hz, and (3) 3650-4000 Hz.

  With this setting, as shown in Fig. 4, the original low frequency component (70-3150Hz) is left unchanged and only the high frequency component (3150-7700Hz) is 3150- for the input sound within the band of 70-7700Hz. Frequency compressed to 4000Hz band.

  The power envelope extraction unit 9 calculates a temporal change in the power of the output signal of the second bandpass filter 8 (3150-4000 Hz). Here, this power envelope is expressed as a moving average of a time window of 1 ms, and the square root is calculated. Incidentally, this square rooting is performed to convert power into an amplitude dimension in order to perform multiplication with a code output from a code extraction unit 11 described later.

  Next, the output signal of the harmonic extraction filter 10 (3150-3400 Hz) is converted into a sign of +1 if the instantaneous amplitude value is positive, 0 if it is zero, and -1 if it is negative. Multiply by the output of the power envelope extractor 9. Further, a band of 3150-3400 Hz is extracted from the output by the compression band filter 12.

  With the above operation, the power change information contained in 3150-4000 Hz of the input sound is compressed (mapped) to 3150-3400 Hz while maintaining harmonic and non-harmonic properties.

  Perform the same operation to compress the power change information contained in the input sound 4000-5300 Hz to 3400-3650 Hz while maintaining the harmonic and non-harmonic properties of the mapping destination, and to 5300-7700 Hz. The included power change information is compressed to 3650-4000 Hz while maintaining the harmonic and non-harmonic properties of the mapping destination. Thereafter, each output (compressed component) and the output of the first band-pass filter are added by the signal adder 6 to generate a frequency-compressed sound.

  In the case of the present embodiment, the pass frequency bands of the harmonic extraction filter 10 and the compression band filter 12 are the same. For this reason, the harmonic information extracted by the code extraction unit 11 completely coincides with the harmonic information of the frequency band to be frequency-compressed, and a more natural frequency-compressed sound can be generated.

  FIG. 8 is a block diagram of a system according to the third embodiment of the present invention. The input band limiting unit 1 includes a first band filter 7 that passes a predetermined frequency component of input speech as it is, and the lower limit frequency is A second band filter 8 that is set to a frequency value equal to or higher than the upper limit frequency of the first band filter 7 and passes the input signal, and a power envelope that extracts the power envelope signal of the output signal of the second band filter 8 An input in-band power calculation unit 2 including an extraction unit 9, a frequency band in which the frequency component of the input signal is compressed as a pass frequency band, an input signal compression band filter 16 that passes the input signal, and the input signal compression Mapping destination in-band power calculation comprising a power envelope extractor 9 for extracting the power envelope signal of the output signal of the band filter 16 15, a power ratio calculation unit 17 that calculates a ratio of output signals of the input band power calculation unit 2 and the mapping destination band power calculation unit 15, the input signal compression band filter 16, and the power ratio calculation unit A power ratio multiplier 18 that multiplies 17 output signals; and a signal adder 6 that adds the output of the input band limiter 1 and the output of the power ratio multiplier 18.

    In the present embodiment, the output signal of the second band filter 7 is compressed within the band of the input signal compression band filter 16 which is narrower than the second band filter 7, and the compressed signal is converted into the first signal. By adding the output of one band filter 7 and the signal adder 6, a sound in which a sound component in a high frequency region is compressed to a low frequency region is generated.

  For compression, the power envelope extraction unit 9 extracts the power envelope (temporal change in power) of the output of the second band filter 8 of the input in-band power calculation unit 2 and the mapped-destination in-band power calculation unit 15. The power envelope extraction unit 9 extracts the power envelope of the output of the input signal compression band filter 16.

  The ratio of the output signal of the input band power calculation unit 2 and the output signal of the mapping destination band power calculation unit 15 is calculated by the power ratio calculation unit 17, and the result and the output signal of the input signal compression band filter 16 are multiplied by the power ratio. Multiply by unit 18. As a result, it is possible to generate frequency-compressed speech with high clarity and naturalness while maintaining the harmonic structure of the input sound, and to construct an efficient system with a smaller amount of computation than in the first and second embodiments. can do.

  FIG. 9 shows, as a system block diagram in the fourth embodiment of the present invention, the input band power calculation unit 2, the mapping destination band power calculation unit 15, the power ratio calculation unit 17, and the power ratio multiplication unit. And the pass band of the second band filter 8 in each input band power calculation unit 2 does not overlap, and the input signal compression band filter 16 passes in each mapping destination band power calculation unit 15. A configuration in which the bands do not overlap is shown.

  In the fourth embodiment, by dividing the high-frequency component of the input sound into a plurality of bands, and compressing and adding the respective in-band signals, it is possible to compress a wider band component into a narrow band. It is possible.

In FIG. 10, in the fourth embodiment, each of the input band power calculation unit 2, the mapping destination band power calculation unit 15, the power ratio calculation unit 17, and the power ratio multiplication unit 18 is provided. The signal aspect in the case is schematically shown. Here, the numbers shown in bold in the figure are the symbols of the corresponding components.
Hereinafter, the present embodiment will be described more specifically with reference to FIG.

  In this embodiment, the boundary frequency (lower limit frequency−upper limit frequency) of the first bandpass filter 7 is 70-3150 Hz, and the boundary frequency of the second bandpass filter 8 is (1) 3150-4000 Hz, (2) 4000- The boundary frequencies of the input signal compression band filter 16 are (1) 3150-3400 Hz, (2) 3400-3650 Hz, and (3) 3650-4000 Hz.

  With this setting, as shown in Fig. 4, the original low frequency component (70-3150Hz) is left unchanged and only the high frequency component (3150-7700Hz) is 3150- for the input sound within the band of 70-7700Hz. Frequency compressed to 4000Hz band.

    The power envelope extractor 9 calculates a temporal change in the power of the output signal of the second bandpass filter 8 (3150-4000 Hz). Here, this temporal change in power is averaged over a 1 ms time window, and the square root is defined as a power envelope a1.

  Next, the power envelope extractor 9 calculates the temporal change of the power of the output signal of the input signal compression band filter 16 (3150-3400 Hz) as the square root of the moving average of the 1 ms time window, and the power envelope b1 To do. A ratio a1 / b1 of the output signal of the second bandpass filter 8 (3150-4000 Hz) with the power envelope a1 is calculated by the power ratio calculation unit 17.

Further, the power ratio multiplier 18 multiplies the output signal of the input signal compression band filter 16 (3150-3400 Hz) and the output a1 / b1 of the power ratio calculator 17.
With the above operation, the power change information contained in 3150-4000 Hz of the input sound is compressed (mapped) to 3150-3400 Hz while maintaining harmonic and non-harmonic properties.

  Perform the same operation to compress the power change information contained in the input sound 4000-5300 Hz to 3400-3650 Hz while maintaining the harmonic and non-harmonic properties of the mapping destination, and to 5300-7700 Hz. The included power change information is compressed to 3650-4000 Hz while maintaining the harmonic and non-harmonic properties of the mapping destination. Thereafter, each output (compressed component) and the output of the first band-pass filter are added by the signal adder 6 to generate a frequency-compressed sound.

  FIG. 5 shows a sound spectrogram of frequency-compressed speech generated by this embodiment. (a) is the original voice, (b) is the voice whose band is limited without performing frequency compression, and (c) is the frequency compressed voice according to this embodiment. It can be seen that high frequency components are compressed while maintaining the harmonic nature of the sound.

  FIG. 6 shows an example of a (a) public broadcasting device, (b) hearing aid, and (c) telephone system to which the present invention is applied. (a) In the public broadcasting device, the announcer speaks into the microphone and performs frequency compression processing on the sound. The frequency-compressed sound is played from a speaker in an existing broadcasting system (broadcasting facility) in a station premises or building.

  In the hearing aid (b), the sound signal collected by the microphone is frequency-compressed, and thereafter, normal hearing processing such as amplification is performed in accordance with the hearing characteristics of the user (deaf person), and the result is output from the earphone. In this embodiment, the hearing aid processing is performed after frequency compression. However, if necessary, the hearing aid processing may be performed before frequency compression.

  In the telephone (c), it is assumed that the telephone 1 and the telephone 2 are making a call. A voice call is performed by performing frequency compression processing on the voice uttered toward the microphone and transmitting the voice. By applying the present invention to each of the devices (a), (b), and (c), it is possible to provide clear and natural sound to the user.

  FIG. 7 shows a system block diagram according to the fifth embodiment of the present invention. The speech enhancement processing device according to the present invention is intended to compress a high frequency component of speech into a low frequency range, but in order to further improve the clarity, in the present embodiment, in the course of compression processing, a consonant, etc. The high-frequency emphasis is performed by selectively amplifying the high-frequency component including the above component.

  Here, an example in which high frequency emphasis is applied to the second embodiment will be described. As shown in FIG. 7, it can be realized by amplifying the output signal of the input band power calculation unit 2 by the amplifier 13 or amplifying the output signal of the output signal limiting unit 5 by the amplifier 14. For example, if the amplifier 13 amplifies the output signal (4000-5300 Hz) of the input in-band power calculation unit 2 by +5 dB and the output signal (5300-7700 Hz) by +10 dB, the compressed sound (3400-3650 Hz) and (3650-4000 Hz) components are emphasized by + 5dB and + 10dB, respectively.

  Furthermore, if the total energy of the output signal of the output signal limiting unit 5 is amplified so as to be equal to the total energy of the output signal of the input in-band power calculation unit 2, a compressed sound having a volume feeling equivalent to the input sound is generated. It is also possible to do.

  Although the second embodiment has been described here, the high frequency components can be selectively amplified in the fourth embodiment as well. In this case, the same effect can be obtained by amplifying the signal input to the power ratio multiplier 18 of the input signal compression band filter 16 and the output signal of the power ratio multiplier 18.

  In the present embodiment, only the speech enhancement by the frequency compression processing has been described. However, when the frequency compression processing of the present invention and other speech enhancement processing methods such as consonant enhancement are used in combination, the effect of the present invention is further enhanced. There is.

  In the present embodiment, the hearing aid, the broadcasting device, and the telephone are mainly described. However, for example, if the frequency compression by the voice enhancement processing device is applied to a loudspeaker such as a megaphone, a natural and clear loud voice can be generated. .

The block diagram of the system in the 1st Embodiment of this invention The block diagram of the system in the 2nd Embodiment of this invention Schematic diagram of signals in the second embodiment of the present invention Conceptual diagram of frequency compression in the second embodiment of the present invention Sound spectrogram An example of a public broadcasting device, hearing aid, and telephone system to which the present invention is applied Block diagram of a system in the fifth embodiment of the present invention The block diagram of the system in the 3rd Embodiment of this invention The block diagram of the system in the 4th Embodiment of this invention Schematic diagram of signals in the fourth embodiment of the present invention

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Input band restriction part, 2 ... Input band power calculation part, 3 ... Harmonic code calculation part, 4 ... In-band signal multiplication part, 5 ... Output signal restriction part, 6 ... Signal addition part, 7 ... 1st 8 ... second band filter, 9 ... power envelope extraction unit, 10 ... harmonic extraction filter, 11 ... code extraction unit, 12 ... compression band filter, 13 ... amplifier, 14 ... amplifier, 15 ... mapping In-band power calculation unit, 16 ... input signal compression band filter, 17 ... power ratio calculation unit, 18 ... power ratio multiplication unit

Claims (10)

In a speech enhancement processing device having a function of compressing a frequency component of an input signal, the speech enhancement processing device includes an input band limiting unit, an input in-band power calculation unit, a harmonic code calculation unit, an in-band signal multiplication unit, The input band limiting unit includes a first band filter that passes a predetermined frequency component of the input signal as it is, and the input in-band power calculating unit has a lower limit frequency. A second band filter that is set to a frequency value equal to or higher than an upper limit frequency of the first band filter and passes the input signal; and a power envelope extraction unit that extracts a power envelope signal of an output signal of the second band filter The harmonic code calculating unit has an arbitrary pass frequency band and allows the input signal to pass therethrough, and the harmonic extraction filter. The in-band signal multiplication unit has a function of multiplying the output signals of the in-band power calculation unit and the harmonic code calculation unit, The output signal limiting unit includes a compression band filter having a frequency band in which a frequency component of the input signal is compressed as a pass frequency band, and the signal adding unit includes the output of the input band limiting unit and the output signal limiting A speech enhancement processing apparatus characterized by having a function of adding the outputs of the units. The speech enhancement processing device according to claim 1, wherein a pass frequency band of the harmonic extraction filter coincides with or is included in the first bandpass filter.   The speech enhancement processing device according to claim 1, wherein a pass frequency band of the harmonic extraction filter coincides with or is included in the compression band filter.   The speech enhancement processing device according to claim 1, 2, or 3, comprising a plurality of the in-band power calculation unit, the harmonic code calculation unit, the in-band signal multiplication unit, and the output signal limiting unit, respectively. A speech enhancement processing device characterized in that the passbands of the second filter in each input band power calculation unit do not overlap and the passbands of the compression band filter in each output signal restriction unit do not overlap. 5. The speech enhancement processing apparatus according to claim 1, wherein the output signal of the input in-band power calculation unit is amplified. 6. The speech enhancement processing apparatus according to claim 1, wherein the output signal of the output signal restriction unit is amplified. In the speech enhancement processing apparatus having a function of compressing the frequency component of the input signal, the speech enhancement processing apparatus includes an input band limiting unit, an input in-band power calculation unit, a mapping destination in-band power calculation unit, and a power ratio calculation unit. The input band limiting unit includes a first band filter that passes a predetermined frequency component of the input signal as it is, and the input band power calculation unit includes a lower limit frequency. Is set to a frequency value greater than or equal to the upper limit frequency of the first band filter, and a second band filter that passes the input signal and a power envelope extraction that extracts a power envelope signal of the output signal of the second band filter The mapping destination in-band power calculation unit uses the frequency band in which the frequency component of the input signal is compressed as a pass frequency band, and An input signal compression band filter that allows a signal to pass through, and a power envelope extraction unit that extracts a power envelope signal of an output signal of the input signal compression band filter, wherein the power ratio calculation unit includes the input band power calculation unit and the A function of calculating the ratio of the output signal of the mapping destination in-band power calculation unit, and the power ratio multiplication unit has a function of multiplying the output signal of the input signal compression band filter and the output signal of the power ratio calculation unit. And the signal adding unit has a function of adding the output of the input band limiting unit and the output of the power ratio multiplying unit. 8. The speech enhancement processing device according to claim 7, wherein the input in-band power calculation unit, the mapped-destination band power calculation unit, the power ratio calculation unit, and the power ratio multiplication unit each have a plurality, A speech enhancement processing apparatus characterized in that the pass bands of the second filter in the power calculation unit do not overlap and the pass bands of the input signal compression band filters in the mapping target in-band power calculation units do not overlap. 9. The speech enhancement processing apparatus according to claim 7, wherein the output signal of the input signal compression band filter is amplified. 10. The speech enhancement processing apparatus according to claim 7, 8 or 9, wherein the output signal of the power ratio multiplication unit is amplified.

Images