CST: VoiceNavigator

VoiceNavigator: voice self-service for contact centers

• Recognition and speech synthesis
• Support of the leading IVR platforms
• Compatibility with open source IP PBX.

General information

VoiceNavigator is the high-tech solution for contact centers intended for creation of the Systems of Voice Self-service (SVSS). VoiceNavigator allows to process automatically calls using technologies of synthesis and speech recognition.

Scope

Corporate system of voice self-service of clients with difficult multilevel menus Help systems of mass information service: answerphones of the schedule of the movement of passenger transport, services of automatic armoring of tickets and services, etc. The systems of voice authentication of users at a request of the personalized or confidential information by phone or via the Internet.

Component structure

The system of voice self-service VoiceNavigator consists of four hi-tech modules: MRCP servers, the module of diktoronezavisimy recognition of the Russian speech, the module of automatic synthesis of the Russian speech and the module of biometric verification on a voice.

VoiceNavigator Server

The MRCP server providing interaction of an IVR system of the client and VoiceNavigator the main communication module organizing interaction of an IVR system of different producers with the server of speech applications and modules of recognition, synthesis and verification connected to it.

VoiceNavigator TTS

the module of automatic synthesis of the Russian speech (TTS — Text-to-Speech) The module provides natural sounding which is reached as at the acoustic level (due to use of the hybrid technology combining achievements allofonny and Unit Selection of approaches), and at the level of the correct interpretation of reductions, numbers, abbreviations, different special text signs taking into account features of grammar of Russian.

VoiceNavigator ASR

The module of recognition of the Russian speech (ASR — Automatic Speech Recognition) It is intended for ensuring the accuracy of speech recognition necessary for work of the systems of voice self-service in real time, the module considers the following factors: level of surrounding noise, transmission channel of the voice information (GSM, ISDN, VoIP, analog communication lines), subjects of a recognizable conversation and feature of Russian.

VoiceNavigator SIV

The module of voice verification (SIV — Speaker Identification Verification) the unique technology allowing the system of voice self-service on the basis of VoiceNavigator "uznavat" of the client on his voice and to personalize further communication in an IVR system.

Scheme of work VoiceNavigator. Source: CST

Characteristics of the MRCP server

The software package STC MRCP Server (MRCP server) supports interaction with the following voice platforms:

• Genesys Voice Platform
• Avaya Voice Platform
• Cisco Unified Contact Center Express
• Siemens OpenScape CC Voice Portal
• FreeSwitch together with UniMRCP library
• Asterisk together with UniMRCP library
• The MRCP server supports the MRCP protocol of version 1.

• The MRCP server supports formats: VoiceXML, SRGS, SSML, NLSML.

• The MRCP server maintains compatibility with Open Source UniMRCP library.

Speech synthesis complex

2019: Announcement of third generation of speech synthesis

CST group (enters an ecosystem of Sberbank) provided on November 21, 2019 third generation of speech synthesis.

As representatives of CST told TAdviser, the updated speech synthesis provides smoothness and expressiveness of reading of any text in target language - it is its main advantage. The high-quality audiosignal is generated by difficult neural network models. In technology there is also a module of prediction of accents in words which are not in the basic dictionary of synthesis yet and also automatic correction of typical spelling errors. Thanks to deep linguistic analysis of the text pronunciation will meet language standards even in complex cases.

On population survey results which was carried out using the specialized systems in the anonymous mode the quality of synthesis got assessment 4.7 points on a 5-mark scale. For comparison the systems of speech synthesis from main competitors and the living person were used.

Besides, for high-quality synthesis expensive servers are not required now. It is possible to use technology by two methods — through a cloud service or having built in the solution. In the automated scenarios of communication with clients the technology allows to interact in an individual order with each subscriber as there are no fixed messages, and any text can be synthesized in the course of a call.

In third generation we implemented a request for naturalness. In the following iteration an opportunity to manage a range of emotions of the synthesized voice will be implemented — the director of research department of CST Kirill Lyovin reported TAdviser.

2012: Characteristics of a complex of speech synthesis of STC TTS

For March, 2012 the software package of speech synthesis of STC TTS has the following functional characteristics:

• support of standard markup languages: SAPI, SSML
• high naturalness of sounding on any text
• two men's and three female voices
• exact intonational execution of the synthesizable speech
• accounting of phonetic, morphological and syntax features of Russian
• correct interpretation of reductions, numbers, abbreviations and special signs
• correct arrangement of accents
• transliteration of the English words
• extensive morfogrammatichesky vocabulary (8 million word forms)
• possibility of creation of the unique synthesized voice "to order"

Characteristics of a complex of speech recognition

The software package of speech recognition has the following functional characteristics:

• support standard and markup languages of Voice XML: SRGS, NLSML
• recognition of the Russian speech regardless of gender and age of the subscriber, without the need for setup for a specific voice
• resistance to surrounding noise and noises in express speaker circuit
• recognition of the Russian speech works with reliability of 97%
• work with the closed grammars (the set option of pronouncing a command)
• work with open grammars (any option of pronouncing command)
• use of data mining technology in the mode of open grammars
• steady work with the communication channels using compression of a voice signal (GSM, VoIP, ISDN)
• possibility of replenishment of the dictionary
• implementation of specific requirements of the customer

Format of synthesizable audiodata

The software package of speech synthesis of STC TTS together with the MRCP server as a part of the product VoiceNavigator supports the following formats of the output audiodata transferred under the RTP protocol:

• G.711 a-law, mono, 8000 Hz
• G.711 μ-law, mono, 8000 Hz

Format of recognizable audiodata

The software package of speech recognition of STC ASR together with the MRCP server as a part of the product VoiceNavigator supports the following formats of the entrance audiodata transferred under the RTP protocol:

• PCM of 16 bits, mono, 8000 Hz
• G.711 a-law, mono, 8000 Hz
• G.711 μ-law, mono, 8000 Hz
• G.729

History of development of a product

2011

Within the affiliate program of Avaya DevConnect the platform for creation of voice services VoiceNavigator of production of Center of Speech Technologies (CST) company is certified at the end of 2011 on compatibility with Avaya Voice Portal - new generation of intelligent communication solutions. The compatibility of VoiceNavigator and Avaya Voice Portal allows to offer different businesses a broad spectrum of solutions which will change methods of client service and will become decisive competitive advantage in the market.

The platform for creation of voice services VoiceNavigator is intended for creation of the systems of voice self-service in contact centers and allows to process automatically telephone calls using the technologies of synthesis and speech recognition developed in CST. After successful testing VoiceNavigator received the certificate of conformity with Avaya Voice Portal r5.1, the platform of the last generation applied to implementation of the systems of voice self-service. The received certificate confirms compliance of VoiceNavigator to the highest international standards in the field of telecommunications.

"Certification of VoiceNavigator on compatibility with the solution Avaya, first of all, will expand possibilities of our customers, - Hitrov Mikhail, the CEO of the Center of speech technologies says. – The innovative solutions of CST and Avaya will provide stability of service quality in contact center and reduction in cost of creation of the services focused on clients".

The macroregional branch "Severo-Zapad" of JSC Rostelecom already managed to appreciate high efficiency and functionality of a system of interactive voice interaction. Since 2010 here based on uniform contact center the free help desk using technologies of synthesis and speech recognition works. The company management (then still JSC SZT) made the decision to scale service of speech recognition of directory service "09"/8-118-09 and to create new voice services in all regions of presence.

"One of conditions of the project on the choice of the solution was orientation to high-performance infrastructure software products and the innovation applied systems, - the Associate director of macroregional branch – the Commercial director of macroregional branch "Severo-Zapad" of JSC Rostelecom Kulazhenkov Andrey told. - For the choice of an optimal solution testing which results showed stable work and full compatibility of the software package VoiceNavigator with the voice Avaya Voice Portal platform was held".

By means of technologies of synthesis and speech recognition of VoiceNavigator more than 1000 clients at the same time will allow to carry out automatic service. As a result of testing in "Rostelecom is the Northwest" the decision to establish the software package VoiceNavigator for development of voice services that is today an event, unprecedented on scale, in the market of voice technologies in the CIS was made.

The center of speech technologies is a Golden partner of Avaya company in the DevConnect program within which participants present solutions for effective joint work of the partner companies. Within the DevConnect program development, testing and promotion of the innovation products compatible to solutions of Avaya is performed.

2015

The platform for creation of voice services VoiceNavigator passed in the summer of 2015 tests in Avaya Test Lab and was certified for sharing of Avaya Aura Experience Portal 7.0.1 – the system of corporate communications for contact centers.

Now the companies working with Avaya AEP 7.0.1 had an opportunity to significantly simplify for itself the solution of such tasks as:

• creation of corporate systems of voice self-service of clients with difficult multilevel menus;
• creation of the help systems of mass information service (answerphones of the schedule of the movement of passenger transport, services of automatic armoring of tickets and services, etc.);
• creation of the systems of voice authentication of users at a request of the personalized or confidential information in channels of remote service.

Thanks to VoiceNavigator the total number of the serviced calls, percent of the serviced calls from the first address and service rate increases. Clients obtain necessary information more simply and quicker, so, the level of their loyalty increases.

"Certification of VoiceNavigator on compatibility with the solution Avaya, first of all, will expand possibilities of our customers — Veresov Ivan, the chief executive of CST says. — The innovative solutions of CST will provide stability of service quality in contact center and reduction in cost of creation of the services focused on clients".