DOI QR코드

DOI QR Code

An ACLMS-MPC Coding Method Integrated with ACFBD-MPC and LMS-MPC at 8kbps bit rate.

8kbps 비트율을 갖는 ACFBD-MPC와 LMS-MPC를 통합한 ACLMS-MPC 부호화 방식

  • Lee, See-woo (Smart Information and Telecommunication Engineering, SANGMYUNG UNIV)
  • Received : 2018.05.02
  • Accepted : 2018.09.18
  • Published : 2018.12.31

Abstract

This paper present an 8kbps ACLMS-MPC(Amplitude Compensation and Least Mean Square - Multi Pulse Coding) coding method integrated with ACFBD-MPC(Amplitude Compensation Frequency Band Division - Multi Pulse Coding) and LMS-MPC(Least Mean Square - Multi Pulse Coding) used V/UV/S(Voiced / Unvoiced / Silence) switching, compensation in a multi-pulses each pitch interval and Unvoiced approximate-synthesis by using specific frequency in order to reduce distortion of synthesis waveform. In integrating several methods, it is important to adjust the bit rate of voiced and unvoiced sound source to 8kbps while reducing the distortion of the speech waveform. In adjusting the bit rate of voiced and unvoiced sound source to 8 kbps, the speech waveform can be synthesized efficiently by restoring the individual pitch intervals using multi pulse in the representative interval. I was implemented that the ACLMS-MPC method and evaluate the SNR of APC-LMS in coding condition in 8kbps. As a result, SNR of ACLMS-MPC was 15.0dB for female voice and 14.3dB for male voice respectively. Therefore, I found that ACLMS-MPC was improved by 0.3dB~1.8dB for male voice and 0.3dB~1.6dB for female voice compared to existing MPC, ACFBD-MPC and LMS-MPC. These methods are expected to be applied to a method of speech coding using sound source in a low bit rate such as a cellular phone or internet phone. In the future, I will study the evaluation of the sound quality of 6.9kbps speech coding method that simultaneously compensation the amplitude and position of multi-pulse source.

본 논문에서는 합성 음성파형의 일그러짐을 제어하기 위하여 V/UV/S(Voiced / Unvoiced / Silence)의 스위칭을 사용하고, 피치구간마다 멀티펄스를 보정하며, 무성자음(Unvoiced)의 근사합성에 특정주파수를 이용하는 ACFBD-MPC(Amplitude Compensation Frequency Band Division - Multi Pulse Coding)와 LMS-MPC(Least Mean Square - Multi Pulse Coding)를 통합한 8kbps ACLMS-MPC(Amplitude Compensation and Least Mean Square - Multi Pulse Coding) 부호화 방식을 제안하고자 한다. 여러 방식을 통합하는데 있어서, 음성파형의 일그러짐을 줄이면서 유성음과 무성음의 비트율을 8kbps로 조정하는 것이 중요하다. 유성음과 무성음의 비트율을 8kbps로 조정함에 있어서, 개별피치를 이용하여 대표구간의 멀티펄스를 피치구간마다 복원함으로서 음성파형을 효율적으로 합성할 수 있다. 8kbps의 부호화 조건에서 ACLMS-MPC 방식을 구현하고 SNR를 평가한 결과, ACLMS-MPC의 SNR는 남자음성에서 15.0dB, 여자음성에서 14.3dB 임을 확인할 수 있었다. 따라서 ACLMS-MPC가 기존의 MPC, ACFBD-MPC, LMS-MPC에 비하여 남자음성에서 0.3dB~1.8dB, 여자음성에서 0.3dB~1.6dB 정도 개선된 것을 알 수 있었다. 이러한 방법들은 셀룰러폰이나 인터넷폰과 같이 낮은 비트율의 음원을 사용하여 음성신호를 부호화하는 방식에 활용할 수 있을 것으로 기대된다. 향후 멀티펄스 음원의 진폭과 위치를 동시에 보정하는 6.9kbps 음성부호화 방식의 음질평가를 수행하고자 한다.

Keywords

OTJBCD_2018_v19n6_1_f0001.png 이미지

(그림 1) ACLMS-MPC (Figure 1) ACLMS-MPC

OTJBCD_2018_v19n6_1_f0002.png 이미지

(그림 2) 멀티펄스의 음원 (Figure 2) Exciting Source of Multi-Pulse

OTJBCD_2018_v19n6_1_f0003.png 이미지

(그림 3) ACLMS-MPC 시스템 (Figure. 3) System of ACLMS-MPC

OTJBCD_2018_v19n6_1_f0004.png 이미지

(그림 4) MPC, ACFBD-MPC, LMS-MPC, ACLMS-MPC의 SNR (Figure 4) SNR of MPC, ACFBD-MPC, LMS-MPC and ACLMS-MPC

(표 1) 부호화 조건 (Table 1) Coding Condition

OTJBCD_2018_v19n6_1_t0001.png 이미지

(표 2) 음성샘플 (Table 2) Speech Sample

OTJBCD_2018_v19n6_1_t0002.png 이미지

(표 3) MPC, ACFBD-MPC, LMS-MPC, ACLMS-MPC의 SNRseg (Table 3) SNRseg of MPC, ACFBD-MPC, LMS-MPC and ACLMS-MPC

OTJBCD_2018_v19n6_1_t0003.png 이미지

References

  1. Selma Ozaydm, Buyurman Baykal: "Matrix quantization and mixed excitation based linear predictive speech coding at very low bit rates", Speech Communication 41, p381-392, 2003, https://doi.org/10.1016/S0167-6393(03)00009-8
  2. K.Krishna, V.L.N.Murty, K.R.Ramakrishnan: "Vector quantization of excitation gains in speech coding", Signal Processing 81,p203-209, 2001, https://doi.org/10.1016/S0165-1684(00)00200-0
  3. Ghaemmaghami, S.,Sridharan, S.: "Very low rate speech coding using temporal decomposition". IEE Electron. Lett. 35(6), p456-457. 1999, https://doi.org/10.1049/el:19990316
  4. LeBlanc, W.P., Bhattacharya,B.,Mahmoud, S.A.: "Efficient search and design procedures for robust multi stage vector quantization of LPC parameters for 4kbps speech coding". IEEE Trans. Speech Audio Process. p373-385. 1993, https://doi.org/10.1109/89.242483
  5. McCree,A.V.,Barnwell, T.P.,: "A mixed excitation LPC vocoder model for low bit rate speech coding", IEEE Trans. Speech Audio Process, p242-250, 1995, https://doi.org/10.1109/89.397089
  6. Feng Huang, Tan Lee, W.Bastiaan Kleijn, Ying-Yee Kong : "A method of speech periodicity enhancement using transform-domain signal decomposition", Speech Communication, p102-112, 2015, https://doi.org/10.1016/j.specom.2014.12.001
  7. SeeWoo Lee: "A Study on ACFBD-MPC in 8kbps", Journal of the Korea Academia-Industrial cooperation Society, Vol. 17, No.7, p49-53, 2016. http://dx.doi.org/10.5762/KAIS.2016.17.7.49
  8. SeeWoo Lee: "A Study on LMS-MPC Method Considering Low Bit Rate", The Journal of Policy & Management, Vol. 10, No. 5, p233-238, 2012, http://dx.doi.org/10.14400/JDPM.2012.10.5.233
  9. SeeWoo Lee:"A Study on 8kbps IP-MPC Method Considering Cellular Phone", The Journal of Korea Knowledge Information Technology Society, Vol.5, No. 5, p149-154, 2010, http://www.kkits.or.kr/bbs/board.php?bo_table=doc2&wr_id=109