Core Technologies

Overview

AppTek has built a comprehensive platform of human language technologies and products that provide coverage for many types of users and opportunities for multilingual solutions across commercial and government markets. AppTek’s integration of speech and text processing capability and cross-leveraging of resources is unique. Applications for its speech and translation technologies are very broad and include: Monitoring of TV, web and radio content; portable and handheld devices with embedded MT and ASR; telephony monitoring; interactive voice response (IVR) telephony solutions; PC-based dictation; and telephone-to-telephone translation, among others.

Also unique is its combination full-fledged, mature rule-based MT approaches with statistical MT platforms. This does not only apply to MT, but also named-entity detection, knowledge/content management and ASR. AppTek offers products that can solve individual needs, such as translating documents, as well as turnkey solutions to address larger needs, such as high volume data processing of various data types in many different languages. The core patented technology is built on complex algorithms, based on extensive linguistic research, acoustic models and recognition techniques, which require less linguistic training data to adapt the technology to new languages, dialects and uses. The underlying technology platform has been developed on state-of-the-art technology and supports all major operating systems – Linux, Windows, MAC OSX, Android, iOS, and Windows Mobile.

Machine Translation (MT)

Machine Translation is the use of computer software to translate text from one natural language into another. This definition accounts for the grammatical structure of each language and uses rules and assumptions to transfer the grammatical structure of the source language (text to be translated) into the target language (translated text).

AppTek’s Machine Translation (MT) technology provides users the ability to accurately translate text including documents of any format, websites and emails into 30 language pairs and numerous dialects. AppTek’s patent pending Hybrid MT platform is a hybridized system of two fully integrated mature engines of Statistical and Rule-based MT, allowing for a faster and more accurate translation and is capable of processing thousands of words per minute. AppTek’s MT applications can run on stand-alone workstations, existing client/server architecture and through AppTek’s web portal for dynamic translation of online content. These systems can also interface with AppTek’s ASR system to produce Speech-to-Speech Machine Translation.

Automatic Speech Recognition (ASR)

AppTek’s Automatic Speech Recognition solutions were developed based on years of research experience on the man-machine interface. The Company’s modules form the basis for individually tailored speech recognition systems that easily integrate with other AppTek solutions. AppTek’s ASR system is a current-generation, self-learning, phoneme/grapheme and speech recognizer. The AppTek ASR system is a set of engines for speech recognition and speech-to-text conversion. The ASR platform works for both broadcast and unscripted, conversational speech including telephony and microphone. Key features of AppTek’s ASR system include: (1) the ability to manage large vocabulary sets, including for mobile devices; (2) an engine that is dialect and colloquial sensitive and adaptive; (3) an approach that creates unique domain acoustics and language models; (4) special domain models that can be adapted on the fly; (5) fast and reliable recognition of continuous speech with high concurrency; and (6) a platform that is speaker independent. The system manages almost any recognition task: from simple chain of numbers recognition and complete call center solutions to information retrieval to full continuous speech in real time.

Product Portfolios

AppTek’s Omnifluent suite of human language technology (HLT) products, enables automated transcription and translation of multilingual audio, video, and text in more than 30 languages and dialects. This cost-efficient solution provides customers with a framework that combines automatic speech recognition technology with AppTek’s patent-pending hybrid machine translation into a single integrated platform. This scalable environment can process broadcast media, video programming, webcasts, social media, telephone conversations and other forms of multimedia content. Omnifluent products are available through software as a service (SaaS), local installation, and subscription delivery models.

  • Omnifluent Media: Omnifluent Media is a complete multimedia platform that provides automated alignment, closed captioning, subtitling, and rich metadata development for media monitoring and analytics. By automating the labor-intensive task of closed captioning and subtitling, media companies can broaden their reach into new audiences without the significant cost or effort of manual transcription or translation. Through AppTek’s automated rich metadata development, media companies can quickly and efficiently improve search indexing, analytics, and search engine optimization strategies.
  • Omnifluent Translate: Omnifluent Translate is a comprehensive multilingual translation platform that automatically translates multiple formats of both text and audio content. The solution provides content localization, globalization and is integrated for foreign media translation.
  • Omnifluent Mobile: Speech-to-speech translation on mobile devices, cross-lingual interview suite (Talk2Me) and speech recognition integrated in an IVR platform.
Other complementary technologies embedded and decoupled include: named-entity extraction/identification, foreign name matching, knowledge/content management, automated language identification for text and speech (audio).