|
|
| RESEARCH TOPICS (01/01/2009) |
 Interests
The work of Akagi Lab. is speech signal processing and modeling the speech
perception mechanism of humans. Speech is the most natural means of communication
between humans. Therefore, it is believed that modeling these compensation
mechanisms is able to realize the high-performance speech processingsystems.
Research Areas
Fig. 1 shows the principle procedures of speech communication (production
and perception). Akagi laboratory .is mainly focusing on the topics indicated
by the red blocks, such as speech production, speech communication in real
environments, and speech perception. In our research, not only engineering
(digital signal processing) but also knowledge of physiology and/or psychology
is required as shown in Fig. 2.
Fig. 1 Principle of speech communication
Fig. 2 Related research fields
Research Issues
In Figure 3;
Production: Through modeling the speech production mechanisms, the research is aimed
to synthesize the natural speech. The research includes: to investigate
the relationship between speech spectrum and shape of vocal tract, to synthesize
natural speech with non-linguistic information (e.g., individuality and
emotion), to synthesize singing voice.
Perception: Through modeling the perception mechanisms of human beings, the research
is aimed to recognize speech in real-world conditions. Specifically, the
research includes: to realize cock-tail-party effect, to enhance speech.
Fig. 3 View of research topics
1. Non-linguistic Information 
1-1 Singing Voice
1-2 Speaker Indivisuality
1-3 Emotional Speech
1-4 Voice Conversion
1-5 Speech Coding
2. Noise Reduction
2-1 Microphone Array
2-2 F0 Extraction
2-3 De-reverberation
2-4 Bone-conducted Speech
2-5 Speech Recognition
2-6 DOA
3. Cocktail-party Effect Modeling
3-1 Sound Segregation
3-2 Privacy Protection
3-3 Noisy Sound Perception
4. Psychoacoustics
4-1 Auditory Model
4-2 Contextual Effect
4-3 Auditory Filter
4-4 Phase Perception
4-5 Vowel Perception
4-6 Noise Evaluation
5. Physiological Auditory Modeling
6. Abnormal Speech
6-1 Abnormal Speech Perception
6-2 3D Vocal Tract Modeling
7. Interaction between Perception and Production
|
|
|
