| Project/Area Number |
17560341
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Communication/Network engineering
|
|---|
| Research Institution | Tottori University |
|---|
Principal Investigator |
ITOH Yoshio Tottori University, Eng., Professor (70263481)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KINUGASA Yasutomo Matsue College of Technology, Eng, Assistant Professor (40300610)
|
|---|
| Project Period (FY) |
2005 – 2007
|
| Project Status |
Completed (Fiscal Year 2007)
|
| Budget Amount *help |
¥1,950,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥150,000)
Fiscal Year 2007: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2006: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 2005: ¥800,000 (Direct Cost: ¥800,000)
|
|---|
| Keywords | noise reduction / voice / wideband noise / sinusoidal noise / adaptive filter / fundamental period detector / 適応線スペクトル強調器 / 線形予測器 / 逆フィルタ |
|---|
| Research Abstract |
The mobile phone terminal is almost spreading to all members about the people. In such a situation, the mobile phone might be used in the environment with high noise level. For this case, a high level noise mixes with the transmitted voice, and the clearness of voice is deteriorated on the receiving side. Then, the introduction of the noise reduction system to achieve a high-quality telephone call under the noisy environment is needed.
As a typical noise reduction method, the microphone array and spectrum subtraction (SS) method are well known. In the microphone array, the number of microphones used increases as the number of noise sources increases. On the other hand, a musical noise is generated in the SS method which uses only a single microphone.
In this research, we have proposed a new noise reduction system which can simultaneously reduce the wideband noise and the sinusoidal noise. The proposed method is composed of adaptive line enhancer (ALE), adaptive noise estimation filter (NEF) and adaptive linear predictor (LP). The advantage is not to detect a silent section of voice and to be able to track also the change of the characteristic of the noise. The main study results of this research are as follows.
1. It is necessary to set the de-correlation parameter to 20ms or more in ALE to estimate the sinusoidal noise.
2. It is necessary to set the decorrelation parameter to several milli seconds in ALE that suppress the wideband noise and estimate only the voice.
3. The reduction in the computational complexity became possible by using only the taps of corresponding to the fundamental period of the sinusoidal noise in ALE for the sinusoidal noise estimation.
4. The fundamental period detection of sinusoidal noise under the environment where the voice exists together became possible by using ALE for the fundamental period detector.
|
