Speech Technologies

NLU Speech Recognition

Better IVR customer dialogs with CreaLog speech recognition

Telephone service generally is the primary point of contact with a customer. The faster the connection to a competent person, the more customer acceptance you will get. In addition to this aspect, the "agent" cost factor also plays an important part. Speech recognition enabled IVRs can create a win-win situation with speed and cost savings if used in pre-qualification and self-services.

CreaLog IVR Portals achieve consistently high recognition performance

The state-of-the-art recognition software and the syntax developed by CreaLog facilitate not only the recognition of individual words but also the description of entire sentence contexts. This makes it quick and clear to describe complex input options so that effective dialogs in natural language (NLU) can be implemented.

A large number of highly varied self-service options have already been implemented by CreaLog in many industries. Please download our Voice User Interface Design Data Sheet for details or check our Testimonials.

Features

  • Use of market-leading cutting-edge technology
  • Speech recognition for 80 world languages
  • Syntax analysis for sentence contexts
  • MRCP and VXML support

With barge-in (also called talk-in, talkover or cut-through), utterances by the caller interruptiong prompts can also be recognized. This accelerates dialogs and navigation is faster and freer. 

The CreaLog team helps you to optimize the use of individual speech recognition solutions with Professional Services for process and dialog analysis, vocabulary and syntax design as well as ongoing acceptance and success analyses. 

Download the Speech Processing Data Sheet for more Information.


Text-to-Speech - Speech synthesis

Read out text and database content with CreaLog Speech Synthesis

Provision of voice portals, reading out email, web sites, news, traffic reports or other regularly updated information, implementing intelligent personal assistants, facilitating voice output of UMS (unified messaging) – all this is possible with speech synthesis.

Speech Synthesis, also called Text-to-Speech (TTS), is usually used where large volumes of information need to be read aloud or information is frequently updated. It can be used to convert any written text into speech.

Precise knowledge of a language makes it possible to achieve perfect pronunciation; years of development work by various manufacturers make speech synthesis comprehensible and sound as natural as possible today.
The existing text is broken down by the system, using various algorithms, into ever smaller units (sentence, word, syllable etc.) until the lowest level of segmentation is reached. Then the pronunciation of the words is established and the linguistic structure is analyzed.
This analysis is then used to determine the melody and rhythm of speech and pauses. The synthesized speech is then played back using diphones previously recorded by a "real" speaker and using specific dictionaries of exceptions. Verbal communication with machines via voice input and output will soon be the norm. 
The use of automated systems with speech synthesis makes it much easier to access timetables, flight departure and arrival times or event dates. CreaLog offers you state-of-the-art technology from a variety of manufacturers, tailored to your specific needs.

Download the Speech Processing Data Sheet for more Information

Corporate voice

Corporate Voice makes your speech synthesis sound perfectly natural and, in future, exactly in your company's "voice".
Your individual speech synthesis is created from the voice of your company, i.e. the voice in your advertising or your voice on the phone.
This individual speech synthesis can then be integrated seamlessly with existing voice recordings and used in a variety of applications in your company.

Results

  • A consistent acoustic (corporate) experience
  • Great flexibility with new announcements
  • Low costs for regular changes

Features

  • multilingual
  • multi-channel
  • supports SAPI and VoiceXML industry standards
  • user-definable pronunciation dictionaries
  • variable synthesis parameters

CreaLog Speech Analytics

Analyze the content of conversations

Once a conversation or a voicemail is transcribed and stored as text (Speech-to-Text) it is ready for in-depth analysis. CreaLogs AI fueled Speech Analytics automatically ascertains information about the structure, content and characteristics of conversations as well as characteristics of the speaker.

This information is predominantly used for quality assurance, process optimization and in business intelligence.

On the one hand it is possible to search through existing voice data for specific information and to prepare statistics. On the other hand, it is possible to react immediately to particular events in live operation by means of the Voice Portal (i.e. IVR) or the supervisor.

Analyzing the dialog structure

Apart from a call's actual content, the dialog's structure can also provide important insights. For example, CreaLog Speech Analytics can help determine how speech and silence are distributed within a call dialog. Are there conversation lulls for any length of time? Does one dialog partner interrupt the other one? Is there double talk in the dialog? Answers to these questions also provide relevant information about the customer service quality in the call center, room for improvement in process flows and possible education or training needs of individual employees.

Automatically analyze, categorize and evaluate recorded conversations in Service Centers, or for legal and compliance reasons 

  • Determination of speaker characteristics e.g. age and gender
  • Emotion detection
  • Speaker identification
  • Language identification
  • Key word (Topic) spotting

Download our Speech Analytics Data Sheet for more Information


Voice Biometrics

High security with CreaLog Speaker Identification and Verification - VoiceID

  • Fast, easy and available everywhere – user authentication by telephone
  • Enhanced security through direct authentication – reduced fraud opportunities
  • Cost effective and flexibly deployable, no special hardware required
  • Easy, intuitive handling by the user “Your voice – your password”
  • High reliability with market leading technology
  • For banks, insurance companies, user helpdesk and many other lines of business

The human voice is unique. No two voices in the world are hundred percent alike.

They vary according to physical features like size and form of the larynx and the length and frequency of the vocal chord.

Apart from that, acquired features like dialect, accent, speech disorders or language habits play an important role. This uniqueness is used by SpeakerVerification to carry out a secure and comfortable verification, be it alone or in combination with passwords, passphrases or a PIN.

Download the Speech Processing Data Sheet for more Information

Similar terms

Voice Print, Speaker authentication, speaker legitimation, speaker verification, voice as a password, voice biometrics, sound biometrics, speaker biometrics, speaker identification


Count on our 25 years of expertise

With more than 400 customers from all industries we have a proven record of success. Please check our reference overview or directly download our A1 Telekom Austria, PAYBACK or ITDZ case studies.

Contact

General requests
+49 89 324 656-0
Service Hotline (24/7)
+49 89 324 656-10

Newsletter

Please subscribe to our newsletter here!

How to reach us

Address

CreaLog
Frankfurter Ring 211
80807 Munich
Germany

How to reach us

Please use our Google Maps Entry and park comfortably behind our building on the signposted CreaLog customer parking lots.