Abstract:
모드스위칭에기초하여윈도우시퀀스를처리하는통합음성/오디오부/복호화기가개시된다. 통합음성/오디오부/복호화기는모드스위칭이발생하는경우, 폴딩포인트를기준으로프레임간 오버랩을수행하여부호화하거나또는복호화할수 있다. 통합음성/오디오부/복호화기는부호화또는복호화를수행하기위해상황에따라다른윈도우시퀀스를처리함으로써, 코딩성능을향상시킬수 있다.
Abstract:
본 발명은 혼합 음악 신호에서 주요한 음원들을 분리 및 추출하는 통합적인 방법 및 시스템이 개시된다. 통합 음원 분리 시스템은 시간과 주파수 영역의 특징을 사용하여 복수의 음원이 혼합된 혼합 음악 신호에서 고유한 시간과 주파수 영역의 특징을 가지는 제1 음원을 분리하는 제1 음원 분리부; 및 스테레오 채널 정보를 사용하여 상기 혼합 음악 신호에서 특정 스테레오 음상 위치에 존재하는 제2 음원을 분리하는 제2 음원 분리부를 포함한다.
Abstract:
시간 측면에서 반복되는 리듬 악기의 특성을 이용하여 혼합 신호에서 리듬 악기로 연주된 음원을 분리할 수 있는 음악 음원 분리 장치가 개시된다. 음악 음원 분리 장치는 혼합 신호를 복수의 세그먼트로 분할하는 세그먼트 분할부; 복수의 상기 세그먼트에 NMPCF(Nonnegative Matrix Partial Co-Factorization) 분석을 수행하고, 분석 결과에 따라 복수의 엔티티(entity)행렬들을 획득하는 NMPCF 분석부; 상기 엔티티 행렬간 내적을 계산하여 복수의 상기 세그먼트 별로 분할된 혼합 신호에서 대상 악기 신호를 분리하는 대상 악기 신호 분리부; 및 각각의 상기 세그먼트에서 분리된 대상 악기 신호를 결합하는 분리 신호 결합부를 포함한다. 음원, NMPCF, NMF, 혼합 신호, 데이터베이스.
Abstract:
PURPOSE: An encoding/decoding apparatus of an audio signal and a method thereof are provided to reduce the size of the encoded audio frame by converting a mute section into a flag type of low capacity. CONSTITUTION: In case an audio signal is in non-mute interval, an audio encoding unit(120) encodes the non-mute speech frame. The audio encoding unit generates a non-mute frame. A mute frame generating unit(130) generates a mute frame of a flag type. An audio frame generating unit(140) combines the non-mute frame and the mute frame.
Abstract:
PURPOSE: A known information compressing device for sound source separation and a method thereof are provided to reduce the size of known information without drop of information necessary for sound source separation. CONSTITUTION: A segment division unit(111) divides sound source information into segments wherein the sound source information is based on musical instruments. A compression information generating unit(112) generates compression information by down-mixing the segments. The known information is made of a plurality of entity matrixes. A plurality of frequency information is overlapped with each entity matrix to generate the compression information.
Abstract:
A Unified Speech and Audio Codec (USAC) for adjusting an overlap area of a window based on a transition is provided. To increase an encoding efficiency, encoding may be performed by overlapping relatively long windows. Additionally, when a transition is generated between frames, an overlap area of a window may be reduced based on the transition, thereby preventing a noise from occurring due to the transition.
Abstract:
PURPOSE: An integration sound source separating device and method thereof are provided to efficiently separate various sound sources within mixing sound signal. CONSTITUTION: A first sound source divider(110) divides a first sound source from a mixed sound signal through the feature of a temporal and frequency domain. A second sound source divider(120) divides a second sound source from the mixing sound signal through stereo channel information. A post-processing unit(130) extracts residual component information from the residual sound source information. A combiner(140) combines second sound source and residual component.
Abstract:
PURPOSE: An integration voice/audio coding/decoding apparatus and method thereof are provided to reduce pre-echo which is generated in a transient section by controlling an overlap domain of a window. CONSTITUTION: An integration voice/audio coding/decoding apparatus includes a detection unit, a first encoding unit, a second encoding unit, and a bit stream formatter. The first encoding unit encodes an input signal and detects a second transient section from an encoded result. The second encoding unit controls the length of an overlap domain and encodes the input signal. The bit stream formatter generates bit stream and the input signal.
Abstract:
PURPOSE: A window processing for interconnecting between an MDTC(Modified Discrete Cosine Transform) frame and a heterogeneous frame is provided to satisfy TDAC conditions for the recovery of an original signal. CONSTITUTION: A frame applying unit(601) applies a window to a current frame. A window folding unit(602) performs the folding of the window. A zero setting unit(603) establishes the domain folding the window in zero value at the current frame. A frame extracting unit(604) extracts the frame in which the folded first window is processed. A window generating unit(605) generates the second window of the opposite value and folded window. A signal restoring unit(606) restores the former state the raw signal from frame.