Dae-Lim Choi

Design and creation of Dysarthric Speech Database for development of QoLT software technology

In this paper we will introduce the work of creation of a speech database to develop speech technology for disabled persons, which has been done as part of a national program to help better life for Korean people. We will report about the creation of speech database of a total of 160 persons: prompting items, designs, etc. for the creation of a database which is needed to develop an embedded key-word spotting speech recognition system tailored for the persons disabled in articulation. The created database is being used by the technology development team in the national program to study the phonetic characteristics of the different types of disabled persons, develop the automatic method to assess degrees of disability, investigate the phonetic features of speech of the disabled, and design and implement the software prototype for personal embedded speech recognition systems adapted to the disabled persons.

Speech/music discrimination using Mel-cepstrum modulation energy

In this paper, we propose Mel-cepstrum modulation energy (MCME) as an extension of modulation energy (ME) for a feature to discriminate speech and music data. MCME is extracted from the time trajectory of Mel-frequency cepstral coefficients (MFCC), while ME is based on the spectrum. As cepstral coefficients are mutually uncorrelated, we expect MCME to perform better than ME. To find out the best modulation frequency for MCME, we make experiments with 4 Hz to 20 Hz modulation frequency, and we compare the results with those obtained from the ME and the MFCC based cepstral flux. In the experiments, 8 Hz MCME shows the best discrimination performance, and it yields a discrimination error reduction rate of 71% compared with 4 Hz ME. Compared with the cepstral flux (CF), it shows an error reduction rate of 53%.

Audio-based automatic detection of objectionable contents in noisy conditions using normalized segmental two-dimesional MFCC

The segmental two-dimensional Mel-frequency cepstral coefficient (STDMFCC) feature has been successfully used in recent studies to detect objectionable sounds, which implicitly represent both static and dynamic characteristics of signal. This study now proposes a new normalized STDMFCC to improve the content recognition performance in diverse noisy environments. Two tests were conducted to verify the performance of the proposed feature: First, an objectionable sound recognition test was conducted with 10-second clips to which white noises with diverse signal-to-noise ratios (SNRs) were added. The proposed feature in the test had an average error reduction rate (ERR) of 24.69% with respect to the STDMFCC. Second, a test was conducted based on the soundtrack that contained diverse channel environments and noises. The equal error rate (EER) of the proposed feature was 4.00% compared with 10.33% of STDMFCC, and the ERR was 61.29%.

Implementation of Korean TTS Service on Android OS

Though Android-based smart phones are being released in Korea, Korean TTS engine is not built on them and Google has not announced service or software developer`s kit related to Korean TTS officially. Thus, application developers who want to include Korean TTS capability in their application have difficulties. In this paper, we design and implement Android OS-based Korean TTS system and service. For speed, text preprocessing and synthesis libraries are implemented using Android NDK. By using Java`s thread mechanism and the AudioTrack class, the response time of TTS is minimized. For the test of implemented service, an application that reads incoming SMS is developed. The test shows that synthesized speech are generated in real-time for random sentences. By using the implemented Korean TTS service, Android application developers can transmit information easily through voice. Korean TTS service proposed and implemented in this paper overcomes shortcomings of the existing restrictive synthesis methods and provides the benefit for application developers and users.