Devices related to synthesized speech
Developer: BME TMIT

▪ Profivox development environment for concatenated waveform speech synthesizer (info: Géza Kiss, Gábor Olaszy, Géza Németh)
▪ Hungarian vocalizer for numbers, dates, time, exchange rates, opening hours (pleasant male and female voices) (info: Gábor Olaszy, Csaba Zainkó, Géza Németh)
▪ Technology to check and correct sound boundaries (info: Gábor Olaszy, Mátyás Bartalis, Csaba Zainkó)
▪ Speech database used for synthesizing weather forecasts (info: Csaba Zainkó, Gábor Olaszy)
▪ Vocalizer: news (info: Géza Kiss, Gábor Olaszy)
▪ Part-of-speech analyzer (info: Géza Kiss)
▪ Formant analyzer (info: Tamás Bőhm)
▪ HMM Hungarian speech synthesizer (info: Bálit Tóth, Géza Németh)
▪ Stress assigner, based on linguistic rules (info: Kálmán Abari)
▪ Accentuator program to restore unaccented Hungarian texts (info: Géza Németh, Csaba Zainkó)
▪ Vocalizer: separate syllables (info: Csaba Zainkó)
▪ Vocalizer: separate sounds (info: Csaba Zainkó)

Contact: +36 1 463 3424, +36 1 463 3883

Devices related to speech recognition
Developer: BME-TMIT and AITIA International Inc.
(Info: Péter Mihajlik, Tibor Fegyó)
 

▪ VOXerver - scalable, multichannel recognition application ( for Windows and Linux, phone or broadband speech). The application performs
    ▪ isolated-word command recognition
    ▪ keyword search in live speech
    ▪ continuous speech recognition (large vocabulary) for telephone and microphone input
    ▪ speech recognition in morphologically rich languages
    ▪ audio-video indexing
▪ General WFST (Weighted Finite State Transducer) toolkit. For generating recognition networks, for parsing recognition outputs, for building linguistic models, etc.
▪ General rule-based phonological pronunciation modeling device (supported languages: Hungarian, Italian, German)
▪ Models for speech recognition (linguistic, pronunciation, acoustic) for Hungarian
 
Devices related to speech acoustics and speech recognition
Developer: BME TMIT Laboratory of Speech Acoustics

▪ Language-independent automatic phoneme segmentation tool (info: Klára Vicsi)
▪ Continuous speech recognition development environments (medium vocabulary) (info: Klára Vicsi)
▪ Prosody recognition tool, with automatic detection of word boundaries (info: Klára Vicsi, György Szaszák)
▪ Emotion recognition tool (info: Dávid Sztahó, Klára Vicsi)

Electronic Dictionaries
Developer: MorphoLogic Ltd.
These award-winning dictionaries are electronic versions of well-known print dictionaries. Several coupling of first and second languages can be found, in more or less detail, from general to specialized dictionaries.

Translation programs
Developer: MorphoLogic Ltd.
The MetaMorpho family of translation programs aims at automatic translation of documents and webpages. They translate to English from Hungarian, or to Hungarian from English.

Proofing tools
Developer: MorphoLogic Ltd.
These proofing tools check spelling, hyphenation, correct language use and the listing of synonyms automatically in Hungarian texts.

Language-learning software
Developer: MorphoLogic Ltd.
This easily searchable database of Hungarian morphology and the whole inflexion system, along with the program to highlight the closed [e] sounds makes the teaching and learning of Hungarian considerably less difficult.

Translation support
Developer: Kilgray Translation Technologies
The MemoQ server, MemoQ and the TM Database offer effective translation support.

Chatbot
Developer: Aitia International Inc.
AITIA International, Inc. has been engaged in artificial intelligence research and development for several years. A Chatbot (Chatter robot) is a software component that interacts with visitors, customers and users making use of the achievements of artificial intelligence research, speech processing and other advanced technologies.

Speech controlled phone customer service: VOXenter
Developer: Aitia International Inc.
The speech-controlled phone Customer Service aims at putting calls through quickly and reliably to the required person or department.

Box of Tricks – audio-visual, computer-supported feedback system for speech correction and learning
Box of Tricks is a multisensory speech teaching system, running on IBM PC compatible computers. It can be used to help correct the speech of children with or without hearing problems, or of hearing-impaired adults. In the course of speech therapy, it places most emphatic weight on the creation of the child’s own basis of articulation, the formation, fixing and making automatic the right pronunciation of Hungarian speech sounds. It can also be used to form and practice the basic, general speech characteristics (loudness, pitch, rhythm, intonation, tone) correctly.
The system displays the most important features of a sound during articulation graphically on the screen. This picture is called the speech picture. The speech picture makes it easy to distinguish between incorrect articulation and the correct model. So the child (or adult with a speech problem) is able to learn the correct pronunciation through looking at their own speech picture and comparing it with the speech picture of the correct pronunciation. It is designed so that even relatively small children (4-6 years old) can use it effectively to learn correct speech.
Together with the visual display, the correct speech picture (syllable, word or sentence) can also be heard through a speaker.
Box of Tricks can make the work of surdopedagogues, speech therapists and phoniatricians more effective and less monotonous. The program also enables children to practice alone, at home, through entertaining games.

Fields of application:
▪ speech development of the hearing impaired
▪ help with speech problems (lisping, etc.)
▪ therapy for delayed speech development
▪ rehabilitation of patients with implants
▪ other speech impediments

Both adults and children (as yet only female patients) use it independently or with the direction of their speech therapist.
In addition to practicing more general speech characteristics (loudness, pitch, rhythm, intonation and tone), the system is best for learning and practicing sibilants and vowels.

Demo: http://www.rcs.hu/sc.htm

MONDOM-2000, GOH hearing screening device
Distributor: http://www.nikol.hu
With this small and compact digital device anyone can make hearing screenings. Screening is done by the use of specially structured synthesized words. Especially for children, the regular cheking of speech hearing can be important, as any problems deterring the normal development of speech hearing should come to light as soon as possible. Age-specific speech perception and speech comprehension are the basis of school work, and both of these rely on proper speech hearing. The screening tells the user whether the child’s speech hearing is normal or deficient (and in what degree). The results yielded by MONDOM-2000 might indicate the problem itself, but the exact nature and degree of deficiency must be established medically. This device is not a medical tool: it only offers general information about the state of the child’s hearing. It is perfectly fitted, however, for regular, fast checks of many children, even in the child’s immediate environment (e.g. the family).
No special expertise is needed to operate the device: its function is exactly to help non-professionals recognize any problems with the development of hearing as soon as possible.
The hearing screening device can be used efficiently by teachers in kindergarten and school, speech therapists, and parents.