JP4852584B2 - Prohibited word transmission prevention method, prohibited word transmission prevention telephone, prohibited word transmission prevention server - Google Patents

Description

  The present invention relates to a method for preventing an inadvertent statement made by a user from being transmitted to a destination, a telephone having the function, and a server having the function.

  In recent years, calls using communication lines have come to be used by users of a wide range of age, from children to the elderly, with the spread of mobile phones. Here, when a young child uses a telephone, his / her home address or telephone number may be inadvertently issued. In such a case, the user's personal information leaks unintentionally. In addition, when emotions are struck during a call, there are cases where the other party makes a statement that makes the other party uncomfortable. Therefore, there is a demand for a technique for appropriately managing words that are prohibited from being used during a call.

Under such circumstances, Japanese Patent Application Laid-Open No. 2004-151867 discloses a technique for making a word conspicuous and improving the efficiency of confirmation work when a word prohibited from being used in a call center is spoken.
JP 2007-68044 A

  However, the technology as described above is intended to easily check whether or not a prohibited word has been issued in the confirmation work after the end of the work, and issuing the prohibited word or phrase itself. It cannot be prevented.

  The present invention has been proposed in view of such problems, and an object of the present invention is to provide a method and a telephone that prevent inadvertent transmission of an utterance made by a user in real time.

  The present invention provides the following solutions.

(1) A method for preventing a voice from being transmitted about an inadvertent utterance made by a user,
An audio accumulation step for accepting and temporarily accumulating user-generated audio;
A voice analysis step of recognizing the voice and analyzing it to a word level;
A determination step of determining whether the analyzed word level speech matches a predetermined prohibited word phoneme model;
A time specifying step for specifying a portion determined to match the prohibited word phoneme model in the speech uttered by the user;
A replacing step of replacing the sound of the identified portion with a dummy sound;
A sending step for sending a voice including the replaced part;
A method comprising the steps of:

  According to the method described in (1), in the determination step, it is determined whether or not the word level speech, which is the recognition result of the temporarily accumulated speech, matches a predetermined prohibited phoneme model. If it is determined that they match, the temporarily accumulated voice is transmitted after the replacement step replaces the portion specified by the time specification step with a dummy sound (for example, a beep). Sent by step. Thereby, even when the user makes an inadvertent utterance, the inadvertent utterance can be replaced with a dummy sound in real time, and an inadvertent utterance can be prevented from being transmitted.

(2) In the method according to (1),
A text conversion step of converting the analyzed word level speech into text using a language model;
In the determination step, in addition to whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model, whether or not the converted text matches a predetermined prohibited word text. Judgment,
The time specifying step specifies a portion determined to match by the determination step.

  According to the method described in (2), in the determination step, it is possible to determine whether it is an inadvertent utterance using text converted using a language model in addition to determination using speech. . Thereby, compared with the method of (1) description, a precision can be improved more.

(3) In the method according to (1) or (2),
It has an address book with information showing the user's interpersonal relationship,
The method further comprising the step of determining whether or not to prevent voice transmission of the inadvertent utterance based on information described in the address book.

  (3) According to the method described in the above, it is possible to consider the interpersonal relationship as to whether or not to prevent inadvertent speech.

(4) In the method according to (1) or (2),
It has an address book with information showing the user's interpersonal relationship,
As a category of the inadvertent speech, including at least prohibited words that make personal information and other people uncomfortable,
The determining step refers to the address book and the category, and performs the determination.

  According to the method described in (4), careless speech can be prevented in consideration of the category of casual speech and interpersonal relationships. For example, while sending personal information to an acquaintance, it is possible to prevent the sending of unpleasant speech.

(5) A forbidden word calling prevention telephone for preventing voice transmission about an inadvertent utterance made by a user,
A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
A voice analysis unit that recognizes the voice and analyzes it at a word level;
A prohibited word determination unit that determines whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model;
A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
A transmitter that transmits audio including the replaced part;
A telephone for preventing prohibited words from being provided.

(6) In the prohibited word calling prevention telephone set forth in (5),
A speech / text converter for converting the analyzed word level speech into text using a language model;
The prohibited word determination unit determines whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model, and whether or not the converted text matches a predetermined prohibited word text. Determine whether
The prohibited-word outgoing utterance time measuring unit specifies a portion determined to be matched by the prohibited-word determining unit.

  According to the prohibited word transmission preventing telephone set forth in (5) and (6), the same effects as the methods described in (1) and (2) can be obtained.

(7) A prohibited word transmission prevention server that manages voice communication through a communication line and prevents voice transmission of inadvertent speech made by a user,
A calling user determination unit for determining a user who has transmitted voice;
A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
A voice analysis unit that recognizes the voice and analyzes it at a word level;
A prohibited word determination unit that determines whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model for the user determined by the calling user determination unit;
A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
A transmitter that transmits audio including the replaced part;
A server for preventing prohibited words from being provided.

  According to the prohibited word transmission preventing server described in (7), the same effect as the method described in (1) can be obtained.

  ADVANTAGE OF THE INVENTION According to this invention, the transmission of the careless utterance which the user uttered can be prevented in real time.

[First Embodiment]
With reference to FIG. 1 to FIG. 5, a prohibited word transmission preventing method and a prohibited word transmission preventing telephone, which are preferred embodiments of the present invention, will be described. In the present embodiment, a forbidden word transmission prevention telephone is described using a mobile phone in which the prohibited word transmission prevention device 30 is incorporated. However, the prohibited word transmission prevention telephone is not limited to a mobile phone, and is used for a fixed telephone. Is also possible.

[Overall configuration of prohibited-word outgoing call telephone 10]
FIG. 1 is a diagram showing an outline of a prohibited word calling prevention telephone 10 in the present embodiment. The prohibited-word calling prevention telephone 10 includes a microphone 11 that receives input of a voice uttered by the user, a transmission unit 13 that transmits the input voice to the other user through an existing communication line, and a voice uttered by the user. And a prohibited word transmission preventing device 30 having a function of excluding prohibited words that are inadvertent statements. Although not shown, the mobile phone 10 is also provided with an A / D conversion unit that converts voice uttered by the user into a digital signal.

  The “prohibited words” include, for example, personal information such as addresses, telephone numbers, birthdays, and words that make the other person uncomfortable, such as predetermined prohibited words such as obscene words, insults, and discrimination words. . Personal information and prohibited words are examples of prohibited words, and the types of prohibited words are not limited to personal information and prohibited words.

  The prohibited word transmission preventing device 30 includes a speech temporary storage unit 31, a voice analysis unit 32, a prohibited word speech determination unit 33, a prohibited word phoneme model 34, a prohibited word speech utterance time measurement unit 35, and a prohibited word replacement unit. 36.

  The temporary sound storage unit 31 temporarily stores digitized sound. The voice analysis unit 32 analyzes the accumulated voice by continuous voice recognition. The outline of the voice analysis unit 32 will be described later with reference to FIG. The prohibited word speech determination unit 33 determines whether the result of speech recognition matches the phoneme model listed in the prohibited word phoneme model 34 (see FIG. 3). That is, the prohibited word speech determination unit 33 determines whether or not it is a prohibited word as it is. The prohibited word speech utterance time measurement unit 35 specifies the speech utterance time of the speech determined to be a prohibited word, that is, the portion corresponding to the prohibited word in the accumulated speech. The prohibited word replacement unit 36 replaces the sound corresponding to the specified prohibited word with a dummy sound.

[Outline of Speech Analysis Unit 32]
With reference to FIG. 2, the continuous speech recognition by the speech analysis unit 32 will be described. This technology is a conventionally known technology. For example, “Netto Technology Dismantling New Book 5 Image / Sound Processing Technology (Sadaaki Furui, Yoshinori Sakai, Denpa Shimbun Co., Ltd., published on January 25, 2004, first edition)” In the following steps S1 to S4 in the figure, only the outline will be described.

  S1: The stored digital audio is divided into a plurality of frames at a predetermined time interval (2 ms to 4 ms), and a power component is extracted from each divided frame. A frame in which the extracted power component exceeds a predetermined threshold is detected as a voice section.

  S2: The sound spectrum of the detected speech section is Fourier transformed to extract an acoustic feature vector.

  S3: Referring to the extracted acoustic feature vector and phoneme model, continuous phoneme recognition is performed for each frame, and the phoneme level is recognized. Note that a hidden Markov model (HMM) can be applied as the phoneme model.

  S4: The speech recognized up to the phoneme level is analyzed at the word level with reference to a word dictionary that models the pronunciation of each word.

[Forbidden Phoneme Model 34]
FIG. 3 is a diagram illustrating an example of the prohibited word phoneme model 34.

  The prohibited word phoneme model 34 stores a prohibited word phoneme model for each prohibited word category. For example, the personal information of the prohibited word category stores "tokyojoto ... (Tokyo ...)" indicating an address. If the analysis of the word level output as a result of speech recognition by the speech analysis unit 32 matches the prohibited word phoneme model stored in the prohibited word phoneme model 34, it is determined to be a prohibited word.

[Main hardware configuration of the prohibited-word calling prevention telephone 10]
FIG. 4 is a schematic diagram showing the main hardware configuration of the prohibited-word outgoing call prevention telephone 10.

  The prohibited word calling prevention telephone 10 has a bus 22. A CPU (Central Processing Unit) 23, a RAM (Random Access Memory) 24, a ROM (Read Only Memory) 25, an input device 26, a communication device 27, and a display device 28 are connected to the bus 22.

  The CPU 23 implements various functions in cooperation with each hardware by appropriately reading and executing various programs stored in the ROM 25. The RAM 24 is a local memory used for program execution. The input device 26 accepts input by the user, and may include a microphone, a keyboard, and the like. The display device 28 displays a screen for accepting data input to the user, and includes a display device such as a liquid crystal display device (LCD).

[Basic operation of the prohibited-phone call prevention telephone 10]
FIG. 5 is a diagram showing the details of the processing flow (steps S201 to S207) in the prohibited-word call prevention telephone 10.

  S201: The CPU 23 stores digital voice data obtained by digitizing voice uttered by the user in the RAM 24 (temporary voice storage unit 31).

  S202: The CPU 23 (voice analysis unit 32) performs voice recognition of the digital voice data stored in the RAM 24. The outline of voice recognition is as described in FIG. By this process, the voice uttered by the user can be recognized at the word level.

  S203, S204: The CPU 23 (prohibited word sound determination unit 33) determines whether or not the sound for each recognized word matches a predetermined prohibited word phoneme model. That is, it is determined whether or not prohibited words including personal information and prohibited words are included in the voice uttered by the user.

  S205: As a result of the determination, if a prohibited word is included, the CPU 23 (prohibited word speech utterance time measuring unit 35) measures the utterance time of the speech determined to be a prohibited word.

  S206: The CPU 23 (prohibited word replacement unit 36) replaces the sound of the portion determined as the prohibited word with a dummy sound.

  S207: The communication device 27 (transmission unit 13) transmits the digital audio data including the replaced dummy sound to the partner user through the communication line.

  Thereby, when a prohibited word is included in the voice during a call, the prohibited word can be replaced with a dummy sound in real time. As a result, it is possible to prevent personal information such as an address and a telephone number from being transmitted to the other user, thereby contributing to crime prevention. Further, even when a prohibited word that makes a partner uncomfortable is accidentally uttered, the prohibited word can be replaced with a dummy sound in real time, so that the partner is not made uncomfortable.

[Second Embodiment]
Next, a prohibited word transmission preventing method and a prohibited word transmission preventing telephone according to a second embodiment of the present invention will be described with reference to FIG. In recent years, telephones are provided with an address book that stores an acquaintance's telephone number, and the second embodiment is characterized in that it uses the address book to determine whether to replace a prohibited word with a dummy sound. Have.
Note that the hardware configuration of the prohibited-word calling prevention telephone 10A is the same as the hardware configuration of the prohibited-word calling preventing telephone 10, and thus the description thereof is omitted.

[Overall configuration of prohibited-word calling prevention telephone 10A]
FIG. 6 (1) is a diagram showing an outline of the prohibited word call preventing telephone 10A in the second embodiment. The prohibited word transmission preventing telephone 10A includes an address book 15 in addition to the prohibited word transmission preventing telephone 10 of the first embodiment. The address book 15 stores information such as the telephone number of the user's acquaintance. The prohibited word transmission preventing device 30 uses information stored in the address book 15 to determine whether to replace the prohibited word with a dummy sound.

[Basic operation of prohibited word calling prevention telephone 10A]
FIG. 6 (2) is a diagram showing details of the processing flow (steps S201 to S207, S211) in the prohibited word call prevention telephone 10A. In the process in the prohibited word transmission preventing telephone 10A, a determination process using an address book is provided in the process between S202 and S203 among the processes in the prohibited word transmission preventing telephone 10 (S211). Hereinafter, a difference from the processing in the prohibited word call prevention telephone 10 will be described.

  S211: When the voice analysis unit 32 performs voice recognition of the digital voice data stored in the temporary voice storage unit 31, the CPU 23 (registration determination unit 41) registers the telephone number of the communication partner user in the address book 15. It is determined whether or not the received telephone number. When this determination is YES, the CPU 23 performs the prohibited word determination (S203) to the voice transmission (S207), and when NO, the CPU 23 performs the voice transmission (S207) without replacing the dummy sound.

  As described above, according to the prohibited word transmission preventing method and the prohibited word transmission preventing telephone 10A according to the present embodiment, the prohibited word transmission preventing function can be operated only when a call is made to an unregistered call destination. The determination process in the address book (S211) is performed after the voice recognition process (S202), but the present invention is not limited to this. For example, it may be performed before the voice recognition process. Thereby, when it is not necessary to operate the prohibited word transmission preventing function, the voice data can be transmitted immediately without performing voice recognition.

  In addition, the prohibited word transmission preventing function is not activated when the destination is registered in the address book, but the present invention is not limited to this. Specifically, the prohibited word category “personal information” may be notified to an acquaintance registered in the address book, but may be problematic if known to a callee not registered in the address book. On the other hand, the prohibited word category “prohibited words” may cause a problem in human relations when used for an acquaintance registered in the address book.

  Therefore, it may be determined whether or not the prohibited word transmission prevention function is activated by referring to the address book registration and the prohibited word category information. In this case, after the prohibited word determination process (S203) is performed, the CPU 23 (registration determination unit 41, prohibited word voice determination unit 33) determines whether the address book registration and prohibited word category information are referred to. It can be realized by doing.

[Third Embodiment]
Next, with reference to FIG. 7 to FIG. 9, a prohibited word call preventing telephone 10B according to the third embodiment will be described. In the above embodiment, whether or not it is a prohibited word is determined by (1) speech (prohibited word phoneme model), but in this embodiment, in addition to (1) determination based on speech, (2) It is characterized by performing determination based on text in consideration of a language model.
Note that the hardware configuration of the prohibited-word call prevention telephone 10B is the same as the hardware configuration of the prohibited-word call prevention telephone 10, and thus the description thereof is omitted.

[Overall configuration of prohibited-word calling prevention telephone 10B ]
FIG. 7 is a diagram showing an outline of a prohibited word call preventing telephone 10B in the third embodiment. In addition, about the structure similar to the prohibited word transmission prevention telephone 10 in the said embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

The prohibited word transmission preventing telephone 10B includes a language model 51, a voice / text conversion unit 52, a prohibited word text determination unit 53, and a prohibited word text list 54 in addition to the configuration of the prohibited word transmission preventing telephone 10.

  The language model 51 is obtained by converting appearance probabilities and connection probabilities for each word into data, including syntax knowledge (grammatical structure), semantic knowledge (relationships and attributes between words), context knowledge (conversation flow), conversation This is a model of general knowledge. Forbidden words (especially prohibited words) may vary depending on utterances and dialects, and it is preferable to use a language model when the speech data of all prohibited words cannot be comprehensively identified in advance.

  In addition, as described in the paper “Post-integration of speech recognition results by multiple topic language models (Ken-ichi Ken)) published at the 2008 Acoustical Society of Japan Spring Conference, using multiple topic language models. By decoding the utterance independently and selecting the hypothesis with the best score for each utterance, higher recognition accuracy can be obtained than when using a single general-purpose language model or topic language model. Therefore, a plurality of topic language models are preferably used for the language model 51.

  The voice / text conversion unit 52 converts the voice analyzed by the voice analysis unit 32 into text using the language model 51. The prohibited word text determination unit 53 determines whether or not the converted text matches the prohibited word text listed in the prohibited word text list 54. The prohibited word text list 54 stores prohibited word text for each prohibited word category, as shown in FIG.

[Basic operation of prohibited-word calling prevention telephone 10B ]
FIG. 9 is a diagram showing the details of the processing flow (steps S201 to S207, S251 to S253) in the prohibited word call prevention telephone 10B . Incidentally, Stopwords originating prevent telephone 10B, in addition to the determination based on (1) voice, (2) where a determination based on a consideration text language model, the determination (S202 to S205) Based on (1) audio Since it is the same as the process of 1st Embodiment, description is abbreviate | omitted. In the following, a description will be given of (2) text-based determination processing in consideration of the language model, which is a characteristic part in the present embodiment.

  S251: The CPU 23 (voice / text converter 52) converts the voice recognized by the voice analyzer 32 into text using the language model 51.

  S252, S253: The CPU 23 (prohibited word text determination unit 53) determines whether or not the converted text matches a predetermined prohibited word text. Then, if it matches the prohibited word text, the processing of S205 and S206 is performed.

As described above, according to the prohibited word transmission preventing telephone 10B in the present embodiment, it is possible to determine a prohibited word using not only speech but also text converted from speech using a language model. Therefore, it is possible to determine whether or not prohibited words are included in the voice uttered by the user with higher accuracy than the determination using only the voice.

  Also in this embodiment, determination using an address book can be performed, and determination using an address book and a prohibited word category can be performed.

(Deformation)
In the above embodiment, the prohibited word transmission preventing device is stored in the telephone owned by the user as an example. However, the prohibited word transmission preventing apparatus is stored in a server in an exchange or an IP telephone for interconnecting telephone lines to form a telephone network. It is good to do.

[Overall Configuration of Prohibited Word Transmission Preventing Exchange / Server 60]
FIG. 10 is a diagram showing an outline of the prohibited word transmission preventing exchange / server 60 in which the prohibited word transmission preventing device 30C is stored. In addition, about the structure similar to the prohibited word transmission prevention telephone 10 in the said embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted. The prohibited word transmission preventing exchange / server 60 includes a communication unit for communicating with the user's telephone, but the illustration is omitted.

  The prohibited word transmission preventing device 30C of the prohibited word transmission preventing exchange / server 60 includes a transmission user determining unit 61 and a prohibited word phoneme model DB 34C in addition to the configuration of the prohibited word transmission preventing device 30. The calling user determination unit 61 identifies the user who has sent the voice. The prohibited word phoneme model DB 34C stores prohibited word phoneme models for each prohibited word category in association with each user.

  The prohibited word transmission prevention exchange / server 60 may also perform (2) determination based on text in consideration of the language model in addition to (1) determination based on speech, and an address book for each user. It is good also as performing determination using an address book and an address book and a prohibited word category.

  As mentioned above, although embodiment of this invention was described, this invention is not restricted to embodiment mentioned above. For example, it can be used as a means for preventing performers from inadvertently speaking broadcast-prohibited terms in live TV and radio broadcasts.

It is a figure which shows the function structure of the prohibited word transmission prevention telephone in embodiment of this invention. It is a schematic diagram of the audio | voice analysis part in the said embodiment. It is a figure which shows the prohibited word phoneme model in the said embodiment. It is a figure which shows the hardware constitutions of the prohibited word transmission prevention telephone in the said embodiment. It is a flowchart of the process of the prohibited word transmission prevention telephone in the said embodiment. It is a figure which shows the function structure and process of a prohibited word transmission prevention telephone in 2nd Embodiment. It is a figure which shows the function structure of the prohibited word transmission prevention telephone in 3rd Embodiment. It is a figure which shows the prohibited word text list in the said embodiment. It is a flowchart of the process of the prohibited word transmission prevention telephone in the said embodiment. It is a figure which shows the function structure of a prohibited word transmission prevention switch and a server.

Explanation of symbols

10 Prohibited Word Transmission Preventing Phone 30 Prohibited Word Transmission Preventing Device 31 Temporary Voice Storage Unit 32 Speech Analysis Unit 33 Prohibited Word Speech Judgment Unit 34 Prohibited Word Phoneme Model 35 Prohibited Word Speech Speaking Time Measurement Unit 36 Prohibited Word Replacement Unit

Claims (7)

A telephone provided with an address book that stores a telephone number of a user's acquaintance is a method for preventing outgoing voices about inadvertent utterances made by the user,
An audio accumulation step for accepting and temporarily accumulating user-generated audio;
A voice analysis step of recognizing the voice and analyzing it to a word level;
An inadvertent determination step for determining whether to prevent the transmission of voice about the inadvertent speech based on the information described in the address book;
As a result of the determination, it is determined whether or not the analyzed word level speech matches a predetermined prohibited phoneme phoneme model on condition that it is determined to prevent transmission of speech for the inadvertent speech. A prohibited word judging step for judging;
A time specifying step for specifying a portion determined to match the prohibited word phoneme model in the speech uttered by the user;
A replacing step of replacing the sound of the identified portion with a dummy sound;
A sending step for sending a voice including the replaced part;
A method comprising the steps of: A telephone provided with an address book that stores a telephone number of a user's acquaintance is a method for preventing outgoing voices about inadvertent utterances made by the user,
As the category of careless speech, at least personal information and prohibited words are included,
An audio accumulation step for accepting and temporarily accumulating user-generated audio;
A voice analysis step of recognizing the voice and analyzing it to a word level;
A prohibited word determination step of determining whether or not the analyzed word level speech matches the predetermined prohibited word phoneme model with reference to the address book and the category ;
A time specifying step for specifying a portion determined to match the prohibited word phoneme model in the speech uttered by the user;
A replacement step of replacing the sound of the identified part with a dummy sound;
A sending step for sending a voice including the replaced part;
A method comprising the steps of: The method according to claim 1 or 2, wherein
A text conversion step of converting the analyzed word level speech into text using a language model;
In the prohibited word determination step, in addition to whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model, whether or not the converted text matches a predetermined prohibited word text Determine whether
The time specifying step specifies a portion determined to match by the prohibited word determining step. This is a prohibited word calling prevention phone that prevents the sending of voices about inadvertent utterances made by users,
An address book storage unit for storing an address book for storing telephone numbers of acquaintances of users;
A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
A voice analysis unit that recognizes the voice and analyzes it at a word level;
An inadvertent determination unit that determines whether or not to prevent transmission of voice for the inadvertent utterance based on the information described in the address book;
As a result of the determination, it is determined whether or not the analyzed word level speech matches a predetermined prohibited phoneme phoneme model on condition that it is determined to prevent transmission of speech for the inadvertent speech. A prohibited word determination unit for determining;
A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
A transmitter that transmits audio including the replaced part;
A telephone for preventing prohibited words from being provided.   This is a prohibited word calling prevention phone that prevents the sending of voices about inadvertent utterances made by users,
  An address book storage unit for storing an address book for storing telephone numbers of acquaintances of users;
  As the category of inadvertent speech, a prohibited word storage unit including at least personal information and prohibited words,
  A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
  A voice analysis unit that recognizes the voice and analyzes it at a word level;
  A prohibited word determination unit that determines whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model with reference to the address book and the category;
  A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
  A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
  A transmitter that transmits audio including the replaced part;
  A telephone for preventing prohibited words from being provided. A prohibited word transmission prevention server that manages voice communication via a communication line and prevents voice transmission of inadvertent speech made by a user;
An address book storage unit for storing an address book for storing telephone numbers of acquaintances of users;
A calling user determination unit for determining a user who has transmitted voice;
A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
A voice analysis unit that recognizes the voice and analyzes it at a word level;
An inadvertent determination unit that determines whether or not to prevent transmission of voice for the inadvertent utterance based on the information described in the address book;
As a result of the determination, the analyzed word level speech is determined in advance for the user determined by the transmitting user determination unit on the condition that it is determined to prevent transmission of the voice for the inadvertent speech. A prohibited word determination unit that determines whether or not the prohibited word phoneme model matches,
A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
A transmitter that transmits audio including the replaced part;
A server for preventing prohibited words from being provided.   A prohibited word transmission prevention server that manages voice communication via a communication line and prevents voice transmission of inadvertent speech made by a user;
  An address book storage unit for storing an address book for storing telephone numbers of acquaintances of users;
  As the category of inadvertent speech, a prohibited word storage unit including at least personal information and prohibited words,
  A calling user determination unit for determining a user who has transmitted voice;
  A voice temporary storage unit that receives and temporarily stores the voice uttered by the user;
  A voice analysis unit that recognizes the voice and analyzes it at a word level;
  A prohibited word determination unit that determines whether or not the analyzed word level speech matches a predetermined prohibited word phoneme model with reference to the address book and the category;
  A prohibited word speech utterance time measuring unit that identifies a portion determined to match the prohibited word phoneme model of the speech uttered by the user;
  A prohibited word replacement unit that replaces the sound of the identified part with a dummy sound;
  A transmitter that transmits audio including the replaced part;
  A server for preventing prohibited words from being provided.

Images