2014 Fiscal Year Final Research Report
speech enhancement using parameter estimation between frequency subbands
| Project/Area Number |
24656239
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Communication/Network engineering
|
|---|
| Research Institution | Kitami Institute of Technology |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2012-04-01 – 2015-03-31
|
| Keywords | 雑音抑圧 / 音声強調 / 線スペクトル対 / サブバンド信号 |
|---|
| Outline of Final Research Achievements |
In order to maintain speech communication in noisy environments, a new speech enhancement method using parameter estimation between frequency subbands is developed. This method divides a speech signal into a low subband signal and a high subband signal. While the low subband signal which has higher SNR is enhanced by the conventional method, the high subband signal which has lower SNR is enhanced using information obtained from low subband signal.
The line spectrum pairs (LSPs) are employed as feature parameters of subband signals. The low subband LSPs are transformed into the high subband LSPs by using the relation between two subbands which was learned preliminarily.
|
| Free Research Field |
音声信号処理
|
