Vytautas Rudzionis

Evaluation of methods to combine different speech recognizers

The paper deals with the problem of improving speech recognition by combining outputs of several different recognizers. We are presenting our results obtained by experimenting with different classification methods which are suitable to combine outputs of different speech recognizers. Methods which were evaluated are: k-Nearest neighbors (KNN), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Logistic Regression (LR) and maximum likelihood (ML). Results showed, that highest accuracy (98.16 %) was obtained when k-Nearest neighbors method was used with 15 nearest neighbors. In this case accuracy was increased by 7.78 % compared with best single recognizer result. In our experiments we tried to combine one native (Lithuanian language) and few foreign speech recognizers: Russian, English and two German recognizers. For the adaptation of foreign language speech recognizers we used text transcribing method which is based on formal rules. Our experiments proved, that recognition accuracy improves when few speech recognizers are combined.

DfA implementations for people with vision and hearing disabilities: application and development for information society

This paper presents activities and challenges when implementing information processing technologies for people with hearing and visual impairments. Other than keyboard based input and monitor based for output modalities should be employed for this category of users. More important is that these modalities are crucial element for successful implementation of complex systems designed for disabled people. Some activities carried on in Lithuania implementing applications oriented for disabled people or using speech technologies and targeted to impaired people are presented too.

Advances on the use of the foreign language recognizer

This paper presents our most recent activities trying to adapt the foreign language based speech recognition engine for the recognition of the Lithuanian speech commands. As presented in our earlier papers the speakers of less popular languages (such as the Lithuanian) have several choices: to develop own speech recognition engines or to try adapting the speech recognition models developed and trained for the foreign languages to the task of recognition of their native spoken language. The second approach can lead to the faster implementation of the Lithuanian speech recognition modules into some practical tasks but the proper adaptation and optimization procedures should be found and investigated. This paper presents our activities trying to improve the recognition of Lithuanian voice commands using multiple transcriptions per command and English recognizer.

Analysis of the Possibilities to Adapt the Foreign Language Speech Recognition Engines for the Lithuanian Spoken Commands Recognition

This paper presents our activities trying to adapt the foreign language based speech recognition engine for the recognition of the Lithuanian speech commands. The speakers of less popular languages (such as the Lithuanian) have several choices: to develop own speech recognition engines or to try adapting the speech recognition models developed and trained for the foreign languages to the task of recognition of their native spoken language. The first approach is expensive in time, financial and human resources sense. The second approach can lead to the faster implementation of the Lithuanian speech recognition modules into some practical tasks but the proper adaptation and optimization procedures should be found and investigated. This paper presents some of our efforts trying to adapt the foreign language oriented speech recognition engines for the recognition of the Lithuanian speech commands for the speaker-independent applications. The experimental investigation shows that the promising results could be achieved with relatively modest investments.

On the Adaptation of Foreign Language Speech Recognition Engines for Lithuanian Speech Recognition

This paper presents some of our activities trying to adapt the foreign language based speech recognition engines for the recognition of the Lithuanian speech commands. In recent years several quiet successful speech recognition engines became available for the most popular languages (such as English, French, Spanish, German, etc.). The speakers of a less widely used languages (such as Lithuanian) have several choices: to develop own speech recognition engines or to try adapting speech recognition models developed and trained for the foreign languages to the task of recognition of their native spoken language. First approach is expensive in time, financial and human resources sense. Second approach can lead to faster implementation of Lithuanian speech recognition modules into some practical tasks but proper adaptation and optimization procedures should be found and investigated.

Voice Controlled Interface for the Medical-Pharmaceutical Information System

This paper describes our efforts developing the voice controlled interface for the medical- pharmaceutical information system. It is well known that voice controlled interfaces are of particular importance and may provide significant convenience for healthcare professionals. Many international IT companies provide speech processing by using tools oriented at the physicians or at other professionals working in the related areas. Lithuanian professionals feel a significant lack of speech processing since there are still no attempts to integrate Lithuanian speech processing into medical information systems in this country. The paper presents several IS under development for the use in Lithuanian healthcare institutions. The experimental evaluation shows that it is possible to achieve recognition accuracy (at least 95% of correct recognition) acceptable for the practitioners. The evaluation of system prototypes shows that voice interfaces may lead to the increased convenience of IT systems at the healthcare practitioners.

Recognition of voice commands using adaptation of foreign language speech recognizer via selection of phonetic transcriptions

In recent years various commercial speech recognizers have become available. These recognizers provide the possibility to develop applications incorporating various speech recognition techniques easily and quickly. All of these commercial recognizers are typically targeted to widely spoken languages having large market potential; however, it may be possible to adapt available commercial recognizers for use in environments where less widely spoken languages are used. Since most commercial recognition engines are closed systems the single avenue for the adaptation is to try set ways for the selection of proper phonetic transcription methods between the two languages. This paper deals with the methods to find the phonetic transcriptions for Lithuanian voice commands to be recognized using English speech engines. The experimental evaluation showed that it is possible to find phonetic transcriptions that will enable the recognition of Lithuanian voice commands with recognition accuracy of over 90%.

Control of Assistive Tools Using Voice Interface and Fuzzy Methods

The paper describes voice controlled multimodal assistive system with fuzzy control. Voice commands are often most convenient way to control various assistive tools. For full functionality voice commands need interpretation. The detection of voice boundaries in the long audio recording was implemented. The experimental results of fuzzy based indoor navigation system are presented in this article. The fuzzy control strategy presented bellow works on a given trajectory principle. The position of the device, the distance from the trajectory, orientation and control tasks are evaluated according to visual data. Paper presents the control model and algorithm of a real-life prototype.

Combining different speech recognizers by using CART classifier

This paper presents out results obtained by experimenting with CART classifier which may be used for creating hybrid speech recognition system. We tried to create speech recognition system which is capable of producing more than 95 % accuracy by recognizing 25 drugs and 25 diseases names. For speech recognizers combination we used CART classifier. By using this classifier we obtained 97.58 % average recognition accuracy. Comparing this experiment results with our earlier results we can see that by using CART classifier we obtained 0.58 % lower results than using 15 Nearest neighbors classifier. In this experiment we tried to combine one native (Lithuanian language) and few foreign speech recognizers: Russian, English and two German recognizers. For the adaptation of foreign language speech recognizers we used text transcribing method which is based on formal rules. Our experiments proved that recognition accuracy improves when few speech recognizers are combined and CART classifier is one of the methods that are suitable for combination task.