CN1315082C - Multi-speech prompting method for intelligent network - Google Patents

Abstract

The invention discloses a method for prompting multiple voices in an intelligent network, which is used to determine the playback mode of the voice. When playing a specific voice, read the index number in its "category index area", then by searching the "category index table", determine the playback mode of the voice according to the attribute of the voice in the "category index table". The types of attributes supported by the present invention are greatly increased, and the adaptability and flexibility are greatly enhanced.

Description

Multi-voice Prompt Method for Intelligent Network

The invention relates to a method for realizing multi-voice prompting in intelligent business in the form of intelligent network in telecommunication network.

In the existing intelligent network business, the information interaction between the system and the user is an indispensable process, and it is also a major feature of our intelligent network business. In the process of interaction between the system and the user, it is very necessary for the system to provide relevant prompt information to the user. In order to complete the function of playing prompt information to users, the Chinese INAP (Intelligent Network Application Protocol) regulations stipulate that between SCP (Service Control Point) and SSP (Service Switch Point), between SCP and IP (Intelligent Peripheral), between SSP and The operation structure between SRF (Specialized ResourceFunction). The operation of playing the notification sound involves the identification of the service voice. However, the procedure does not provide other descriptions on the method of voice identification except that the voice identification is described by a 4-byte parameter.

Since the current INAP regulations neither stipulate the voice coding of various services nor give specific coding methods, the existing switch equipment uses numbers without any coding rules to identify voices for various voices, which leads to the following problems:

1. Since there is no unified regulation on the coding value of each business voice like the signal tone of the switch, the switching equipment and intelligent network equipment of each manufacturer give different codes for the same voice content, and the coding does not have attributes. meaning. This is not conducive to the interconnection of equipment from various manufacturers, and is contrary to the original intention of INAP development;

2. When the business needs to play a voice of a certain language type or multiple language types in different playback sequences, the traditional method is to send multiple operations to the playback device in sequence. In terms of implementation, the sending of multiple operations is to send messages between each module through the network or serial port. A large number of operation sending will inevitably increase the transmission load of the network and serial port, resulting in a decrease in system performance;

3. Since INAP regulations do not specify the specific attributes of the current speech coding, it is difficult to query and manage speech resources through the database, and the efficiency is very low.

Therefore, if no coding rules are used for the identification of voice, and only a 32-bit voice identification code (messageID) is used to identify voice resources in various services, then an ID will be any value from 0x00000000 to 0xFFFFFFFF. This solution has no bit field for service description, which makes it very difficult to search and manage voices. Moreover, due to the lack of description bit fields for the playback language and playback sequence, when the service needs to play a piece of voice in a different playback sequence When speaking in a certain language type or multiple language types, the operation needs to be sent separately for each language type, which increases the communication load and leads to a decrease in system performance. In order to solve this problem, Shenzhen Huawei Technology Co., Ltd. once proposed a scheme for numbering the voice identification. This solution enriches the information expression ability of messageID, improves the efficiency of business voice coding, and improves the efficiency of voice data query and network resource usage. In order to compare with the present invention, this scheme is briefly introduced below:

The types of speech coding in this solution include notification tone coding and recording coding. Among them, the notification tone code is used for the interaction between SCP and IP, SCP and SSP, SSP and mailbox, IP and mailbox, and the 32-bit code is divided into multiple bit fields to identify voice content and various attributes (service number, number of languages) , language, language playback sequence, business organizer, business voice serial number). The recording code is used for the interaction between SMS (Service Management System) and IP (Intelligent Peripheral), SMS and SSP (Service Switching Point), SMS and mailbox, and the 32-bit code is divided into multiple bit fields to identify A piece of voice with specific attributes (business number, language, business organizer, business voice serial number).

The encoding formats are as follows:

1) Notification tone coding format

The notification tone encoding format is encoded by four bytes, and the encoding can include 4 different languages and playback sequences.

Table I 8bit 10bit 5bit 9bit business coding field 4 languages sequentially coded bitfields Service sponsor user code bit field Speech code bit field

31 23 13 8 0

In the above table, the lower label is the bit number, and the description of each field is as follows:

1. Service code: Different services correspond to different service codes. The service code range is 0--254. For notification tone codes that can be used for multiple services, the service code value is FF. SCE assigns different codes to all the services generated by it, and the management logic files sent by SCE to SMS include the service codes and whether the services are related to the host user. In this way, the number of services that can be generated in the SCE is limited to 255.

2. Sequential encoding of 4 languages: the encoding format is as follows. Each language has a code value of 0 and 1, 0 indicates that there is no notification sound in this language, and 1 indicates that there is a notification sound in this language. Each language sequence code value can be 01, 10, 11, 00, each of these four code values is taken by Mandarin Chinese, English, and local dialect, so as to determine the sequence codes of other languages, and the default value is 01. Its format is as follows:

Table II 1 bit 2bit 1 bit 2bit 1 bit 2bit 1 bit Mandarin with or without code bit field Mandarin Sequential Codepoint Field English with and without code point fields English sequential code point field Native with or without code point field Vernacular order code point field Code point field for other languages

23 21 20 18 17 15 14 13

Service Sponsor User Code: If the service voice is irrelevant to the service sponsor, its value is 00000. If the service is related to the sponsoring user of this service, the code for the part not related to the sponsoring user is generated by SCE, and its value is 00000; the code related to the sponsoring user is generated by SMS, and the value is not 00000.

Speech code bit field: If the service has nothing to do with the sponsoring user of the service, the speech code is generated by the SCE. If the service is related to the sponsoring user of this service, the codes for the part not related to the sponsoring user will be generated by SCE: the coding range for the part related to the sponsoring user will be specified by SCE (Service Create Environment), and the specific code will be generated by SMS within this range.

A typical notification tone encoding using the above scheme is as follows:

0xABa000CD: Indicates the Chinese voice of the CD number of the AB number service.

0xABb800CD: Indicates the Chinese and English voices of the CD number of the AB number service, first in Chinese and then in English.

0xABd400CD: Indicates the Chinese and English voices of the CD number of the AB number business, first in English and then in Chinese.

2) Recording encoding format

Recording encoding format: four bytes are used to encode, the encoding does not include the playback order of different languages, only the language is indicated in the encoding.

Table three 8bit 1 bit 1 bit 1 bit 1 bit 6bit 5bit 9bit business code Mandarin with or without code English with or without code Local language with or without code Other languages with or without code Need not Business Sponsor User Code speech coding

31 23 22 21 20 19 13 8 0

Wherein, the business coding and the speech coding are the same as the notification tone coding. The 4 languages are coded with only 4 bits, and each bit is 0 or 1 to indicate the language of the recording code. Only one of these four bits can be 1. The recording coding scheme should be consistent with the notification tone coding scheme.

Typical recording codes such as:

0xAB8000CD: The Chinese voice file recording code of the CD voice of the AB business.

0xAB4000CD: The recording code of the English voice file of the CD voice of the AB business.

0xAB2000CD: Local language voice file recording code of the CD voice of the AB business.

0xAB1000CD: Recording and encoding of voice files in other languages for the CD voice of the AB business.

From the above description, we can find that the above solution does solve the above-mentioned problems in standardization, efficiency and management, and basically meets the domestic demand for intelligent network products. However, with the development of intelligent network technology and the expansion of overseas markets for intelligent network products, this coding scheme can no longer meet the requirements of overseas users. Its main shortcomings are as follows:

1. The business is related to voice. At present, during the generation process of the intelligent network service, the language information is directly written into the executable file, that is, the service is related to the language type of the service prompt. In this way, when our business is applied to other countries, when the type of business prompt language changes, the business needs to be regenerated, tested and loaded, which cannot meet the requirements of multiple countries for the same business but different prompt language types;

2. There are few speech language types supported. At present, voice coding can only support up to four language types, which are: Mandarin, English, local dialects, and a language that does not belong to the above-mentioned languages. For overseas customers, it is required to support Swahili, Spanish, French, and Portuguese. Multilingual support cannot meet the requirements;

3. The voice prompt playback order is not flexible. In the current voice identification coding scheme, the playback order of Mandarin, English and local dialects can be flexibly specified, but for other languages that do not belong to the above three languages, the order of playback can only be behind the above three languages. For overseas non-Chinese and non-English-speaking countries, the requirements on the order of voice playback cannot be met;

4. Unable to meet the requirements of multi-currency price data voice playback. In intelligent network services, the system often needs to play special data type data prompts to users, such as the balance on the user's card. In the existing speech coding system, the currency unit and the position of the currency unit change with the type of prompt language. For example, in Chinese, the currency unit is "Yuan", and the currency unit is played at the back of the number . Since the currency unit is determined by the type of prompt language, if you want to broadcast the user’s balance in US dollars in Chinese, it will not be possible. Therefore, it is necessary to expand the existing price playback solution and bring the information of the currency unit when playing the price;

The purpose of the present invention is to solve the above problems and provide a multi-voice prompting method for an intelligent network, which realizes the purpose of supporting more languages and using more flexibly through a new voice identification coding scheme.

In order to achieve the above object, the present invention proposes a multi-voice prompting method in an intelligent network, which is used to determine the playback mode of the voice, and is characterized in that it comprises the following steps: 1) the voice identification of the voice to be played is encoded, and the encoding contains a " Type index" area, determine a "type index number" with the value of each bit in this area; 2) set a "type index table", list each possible "type index" in the above-mentioned "type index" area in the table The attribute of the voice corresponding to the number", the attribute of the voice includes the playback mode; the "category index table" is loaded and set into the SSP by the background, and can also be dynamically filled by the SCP, and the attribute of the voice is stored in the SSP Middle; 3) when needing to play concrete voice, read " kind index number " in " kind index " area, SSP is by searching described " kind index table ", according to " kind index table " described in " kind index table " The attribute of the voice corresponding to the "category index number", determine the playback mode of the voice, and play the voice according to the determined playback mode.

Owing to having adopted above scheme, be not that the attribute of speech is directly written in the speech identification code, but only write an index number in the speech identification code, the attribute of speech is determined with an index table in addition, can greatly like this The bit length of the voice identification code is saved, and the types of attributes supported can be greatly increased. Such as: under the situation that is used for multilingual broadcast, determine voice attribute as with 8 code lengths, can only support four kinds of languages (four languages) with original method, but then can support 258 kinds of languages with this method. On the other hand, after adopting this method, the phonetic attribute itself is not written into the program, but exists in the index table, so when it needs to be modified, it is not necessary to modify the program, but only needs to modify the index table. Greatly enhanced its adaptability and flexibility.

The present invention will be described in further detail below through specific embodiments and in conjunction with the accompanying drawings.

Embodiment one: the intelligent network multi-voice prompting method in this example is to use said method to determine the category of languages and the playback sequence of each language under the situation of multilingual playback, including the following steps: 1) to the voice that needs to be played The code contains the "business code" area, the "language type index" area and the "speech code" area, and the value of each bit in the "language type index" area is used to determine a "language type index number": 2) a "language category index table" is set, and the attribute of the voice corresponding to each possible index number in the above-mentioned "language category index" area is listed in the table, including language category quantity, language number and play order; 3) when When each specific voice needs to be played, read the index number in its "language type index area", and then determine the voice type according to the attribute of the voice in the "language type index table" by searching the "language type index table". The voice playback method, including: the languages to be played and the playback order of each language.

The design of the optimized coding scheme of the present invention can also support the existing coding method of the voice mark, and at the same time, support a new coding scheme of the voice mark through parameter configuration. The new encoding scheme supports up to 256 language categories and supports up to 8 languages at the same time. Its format is as follows:

Table four 8bit 8bit 16bit business code Language Type Index speech coding

In this scheme, the meaning of "service code" is consistent with the existing definition, "speech code" is equal to the existing "service hosting user code" + "speech code", and "language category index" corresponds to the existing "4 language sequence encoding". In order to be compatible with the Chinese national standard, the length is set to 8 bits (if the standard is not considered, it is obvious that the length can be set to other values, such as a longer length can support more combinations, and a smaller length can save word bits number). Because the 8bit capacity is limited, it cannot accommodate the types and playback sequences of 8 languages, so it is used as an index of speech types and playback sequences, and the specific language types and playback sequences are obtained through index lookup tables. The 8-bit index has 256 combinations, which can meet the needs of business voice. In order to be compatible with the old messageID format, the language type and playback order of the old format are also set in this table, and certain techniques are used to achieve compatibility with the old messageID format.

The language type index format is shown in Table 5: new format Language Type Index 8bit old format 4 voices and the first 8 bits of the sequence 8bit

In this table, the first 8 bits of the "4 kinds of voices and their order" in the old format are also set. Due to the special rules of encoding, the first 8 bits of "4 kinds of voices and their order" can uniquely represent the voice type and voice order. In this way, no matter whether the messageID sent by the SCP is in a new encoding format or an old encoding format, the SSP will look up the process according to this table. Compatible with the two encoding formats has been achieved.

Table six Language Type Index Language category and playback order 0b1, 010, 000, 000, 0xa0 mandarin 0b0001, 010, 000, 0x14 English 0b0000001010, 0x02 dialect ob1011100000 0xb8 Mandarin, English 0b1101010000, 0xd4 English, Mandarin 0b1010001100, 0xa3 Mandarin, Dialect 0b1100001010, 0xc2 Dialect, Mandarin 0b0001011100, 0x17 English, dialect 0b0001101010, 0x1a Dialect, English 0b1011101110, 0b1011101000, 0xbb, 0xba Mandarin, English, Dialects 0b1011111100, 0b1011001100, 0xbf, 0xb3 Mandarin, dialect, English 0b1101011110, ob1101011000, 0xd7, 0xd6 English, Mandarin, Dialects 0b1101111010, ob1101001010, 0xde, 0xd2 Dialect, Mandarin, English 0b1111011100, ob1001011100, 0xf7, 0x97 English, Dialect, Mandarin 0b1111101010, ob1001101010, 0xfa, 0x9a Dialect, English, Mandarin

Table 7 is the format of the language type index table on the SSP side. Language Type Index Number of language categories language category 0～255 1～8 language category 1 language category 2 language category 3 Language Category 4 Language Category 5 Language Category 6 Language Category 7 language category 8

The language category can be selected multiple times, and the playback order of each language is determined in turn. The specific content of this table needs to be determined according to specific needs. At the same time, the codes corresponding to the old "4 voices and sequences" (see the above table) should be reserved to support the old messageID format codes, and the new messageID format codes can also use this index. The advantage of this is that SSP, SCP, etc. do not need to make special signs to distinguish messageID formats for compatibility issues, and can be processed uniformly.

The language categories determined in the background and the playback sequence of the selected language categories are set to the foreground with the following data structure for processing.

table eight 8bit 8bit 8bit ····· 8bit Language Type Index Number of language categories language category 1 ····· language category 8

The voice category index is represented by 8 bits, the index range is 0 to 255, the number of voice categories is 8 bits, which indicates the number of language categories played by this index, and 8 bits are used to indicate 1 to 8 language categories corresponding to this index. The language category is represented by 8 bits and supports up to 256 voices. The language category in the table indicates that a maximum of 8 voices can be played sequentially, and these 8 voices are selected from 256 voices.

For example the following structure:

Table nine 0 8 1 3 5 8 9 10 11 253 1 3 1 3 5 7 9 255 255 255

express:

The playback index is 0, and 8 kinds of voices are played, and the sequence of playback is language category 1, 3, 5, 8, 9, 10, 11, and 253.

Playback index 1, play 3 voices, the order of playback is language category 1, 3, 5. Because there are 3 types of sounds to be played, no sounds are played for language categories 7, 9, etc.

The above scheme can achieve at least the following two points:

1. Support for the number of speech language types

The implementation of the above encoding scheme has changed from supporting 4 languages and playing 4 voices at the same time to supporting up to 256 languages and 8 voices playing at the same time, which effectively solves the problem of the number of languages supported. The problem.

2. Support for issues related to business and specific speech language types

After adopting the new coding scheme, what the service fills in during the voice processing is not the specific language type and playback order of the voice, but the language type index of the voice (this index can also be dynamically filled in by the SCP), and the corresponding The number of voices, language types and playback order are translated by the language type index table stored on the SSP, so you can modify the number of prompt languages corresponding to the corresponding index in the language type index table on the SSP without modifying the business. Types and sequences are used to meet the needs of different users for multiple languages, thus solving the problems related to the business and specific prompt voice language types.

Embodiment 2: Support for multi-currency and multi-voice price data playback

In the INAP protocol of ITU-T, the price (Pice) data type is an OCTET STRING (SIZE (4)), and in CINAP it is a BCD code. If such a price range is expressed in binary, only 27 bits are needed, then the remaining The next few bits can be used to indicate the type of currency. Therefore, it is feasible to add currency type information into the encoding of the Price data type, which can not only meet the specifications of ITU_T, but also represent information of multiple currency types, and realize the purpose of configurable currency types.

In order to solve this problem, the method can be used for multi-voice prompts under multi-currency types and multi-language situations, so as to determine the category of currency, the combination rules of currency units and the language used in pronunciation, including the following steps: 1) Encode the voice identification of the voice that needs to be played. The encoding contains a "currency index" area and a "price value" area, and a "currency index number" is determined by the value of each bit in the "currency index" area; 2 ) a "currency index table" is set, and the attribute of the voice corresponding to each possible index number in the above-mentioned "currency index" area is listed in the table, including the combination rules of pronunciation language, currency and currency unit; 3) When it is necessary to play each specific voice, read the index number in its "currency index area", and then determine the The playback method of the voice includes: the currency to be played, the language and the combination rules of each currency unit.

Specifically, in this new multi-currency support scheme, the following encoding method is adopted:

table ten 4bit 28bit currency index area price value area

The 4 bits in the currency index area are used to indicate the currency type of the price value, and a maximum of 16 currency types can be supported. The price value area uses hexadecimal to represent the specific price value, and its value is represented by the smallest unit. The maximum value can reach more than 200 million.

For example: the price value is 1234.56, the language type is Chinese, the currency unit is "Yuan", and the currency index of Chinese "Yuan" is 1, then the value in the issued price price should be: 1001E240 Currency index table format:

Table Eleven language Currency combination rules Chinese RMB A+circle+B+angle+C+point+D+E Chinese Dollar A+dollar+BC+cent+D+E Chinese other currency English RMB . . .

The combination rule ABCD in the table refers to a flag bit, such as 0xFF for A, 0xFE for B, 0xFD for C, 0xFC for D, 0xFB for E, D followed by whether the decimal part should be put together, that is, whether BC is together , the Chinese decimal part has cents, but the dollar does not, only cents. E indicates the order of playing units and numbers, if there is E, it means playing units first, and then numbers.

Other information such as circle, angle, cent, dollar, cent, etc. is a MsgID, which represents the voice.

Adopting the speech coding scheme of the present invention to realize the function of business multilingual voice prompts on the intelligent network not only expands the function of business voice prompt operation and expands the INAP specification, but also brings other benefits, specifically:

1. The new voice coding scheme enhances the functions of intelligent network services and increases the flexibility of services;

2. The new speech coding scheme refines the INAP protocol, proposes a specific method of speech coding, and concretizes the speech coding method that is not clear in the protocol, making contributions to the development of the INAP protocol;

3. The new speech coding scheme expands the INAP operation, proposes the call duration data type, increases the data range processed by the INAP protocol, and contributes to the development of the INAP protocol;

4. The new speech coding scheme enhances the function of INAP. By changing the coding method of the price data type, the information content of the coding is increased, the content of the INAP protocol is enriched, and it contributes to the development of the INAP protocol;

5. The new voice coding scheme accelerates the speed of business development. In the past, to realize multilingual voice prompts (limited to four languages) in the business, it was necessary to use the branch SIB to select according to the language type of the voice in the business, and each playback place required four SIBs to complete. The proposal of the new voice coding scheme, In particular, the language category index can be filled in by the service or by the SCP, so that each playback operation after adopting the new coding scheme can be completed with only one SIB, which greatly reduces the number of playback and playback-related SIBs in the service. number, speeding up business development;

6. The new voice coding scheme accelerates the speed of service transplantation. At present, the intelligent services in the Peru intelligent network project of Shenzhen Huawei Technologies Co., Ltd. are modified and transplanted from the intelligent services in the Kenya intelligent network project. The intelligent service of the Peru intelligent network project is mainly modified or deleted on the basis of the intelligent service of the Kenya intelligent network project. Part of the service process is supported for multi-voice. Due to the irrelevance between the new coded service and the language type of the prompt voice, the transplantation The process is very simple, just modify the configuration file. The adoption of the new coding scheme has greatly accelerated the speed of business migration;

7. The new voice coding scheme solves the problem of multi-currency price playback under multi-voice. The solution to this problem not only meets the needs of overseas intelligent network users, but also provides solutions for the development of intelligent network platforms and services;

In a word, the new service speech coding scheme of the present invention is theoretically feasible and technically feasible. In practical applications, this technology brings great convenience to business development, maintenance and transplantation with its good operability and comprehensibility, and brings huge economic benefits to the promotion of intelligent network intelligent services overseas.

Claims (3)

1, a kind of multi-speech prompting method for intelligent network is characterized in that, comprises the steps:

1) voice identifier of the voice that needs are play is encoded, and contains one " species index " district in the coding, determines one " species index number " with each value in this district;

2) one " species index table " is set, list the attribute of each possible " species index number " pairing voice in above-mentioned " species index " district in the table, the attribute of described voice comprises broadcast mode, described " species index table " loaded by the backstage and is set among the SSP, also can dynamically be filled in by SCP, the property store of described voice is to SSP;

3) in the time need playing to concrete voice, read " species index number " in " species index " district, SSP described by searching " species index table ", attribute according to these corresponding voice of " species index number " described in " species index table ", determine the broadcast mode of these voice, and play this voice according to determined broadcast mode.

2, multi-speech prompting method for intelligent network as claimed in claim 1, it is characterized in that: described " species index " is " category of language index ", described " species index table " is " category of language concordance list ", the broadcast mode of required definite voice is meant the playing sequence of classification and each languages of determining languages under the situation of multilingual broadcast, and described three steps are specific as follows described:

1) voice identifier of the voice that needs are play is encoded, contain " operation code " district, " category of language index " district and " voice coding " district in the coding, determine one " category of language index number " with each value in " category of language index " district;

2) one " category of language concordance list " is set, list the attribute of each possible " category of language index number " pairing voice in above-mentioned " category of language index " district in the table, the attribute of described voice comprises languages categorical measure, languages numbering and playing sequence, described " category of language concordance list " loaded by the backstage and is set among the SSP, also can dynamically be filled in by SCP, the property store of described voice is to SSP;

3) in the time need playing to each concrete voice, read " category of language index number " in its " category of language index " district, SSP is by searching " category of language concordance list " then, attribute according to these corresponding voice of " category of language index number " described in " category of language concordance list ", determine the broadcast mode of these voice, described broadcast mode comprises the languages that should play and the playing sequence of each languages, and plays this voice according to determined speech play mode.

3, multi-speech prompting method for intelligent network as claimed in claim 1 is characterized in that:

Described " species index " is " Currency Type index ", described " species index table " is " Currency Type concordance list ", the broadcast mode of required definite voice is meant under many currency types, multilingual situation, determine the classification of Currency Type, the rule of combination of monetary unit and the used languages of pronouncing, described three steps are specific as follows described:

1) voice identifier of the voice that needs are play is encoded, and contains " Currency Type index " district and " price numerical value " district in the coding, determines one " Currency Type index number " with each value in " Currency Type index " district;

2) one " Currency Type concordance list " is set, list the attribute of each possible " Currency Type index number " pairing voice in above-mentioned " Currency Type index " district in the table, the attribute of described voice comprises the rule of combination of pronounce languages, Currency Type and monetary unit, described " Currency Type concordance list " loaded by the backstage and is set among the SSP, also can dynamically be filled in by SCP, the property store of described voice is to SSP;

3) in the time need playing to each concrete voice, read " Currency Type index number " in its " Currency Type index area ", SSP is by searching " Currency Type concordance list " then, attribute according to " Currency Type index number " described in " Currency Type concordance list " pairing these voice, determine the broadcast mode of these voice, described broadcast mode comprises the rule of combination of Currency Type, languages and each monetary unit that should play, and plays this voice according to determined broadcast mode.