第三代移动通信中的AMR声码器研究
摘要
语音业务一直是移动通信中的主要业务之一，其中最关键的技术就是语音信号的编码与解码。由于无线通信中始终存在着频率资源不足的问题，低速率语音编码成为现代语音技术研究的主要方向。目前4.8kbps～16kbps的编码算法通常都采用CELP算法。
随着第三代无线移动通信技术的发展与成熟，下一代语音编码技术标准也基本定型。3G AMR声码器采用目前在4K以上速率中应用比较成熟的ACELP技术，并使用活动语音探测和舒适噪声等控制方式提高编码器的主观听觉性能和频带的利用率。为了适应未来第三代无线通信的需要，AMR可以在多种工作模式之间切换，速率从4.75kbps到12.2kbps共8种，因而能够适用在各种不同的用途中。AMR声码器采用了广泛使用的ACELP技术。经过长期实践验证，ACELP算法具有编码效率高，算法复杂度低和工作稳定等主要特点。许多新一代的语音编码标准方案中都采用了该算法，例如国际电联的G.723，G.729等。该声码器在未来的移动通信话音中占有重要地位，是第三代语音通信的核心，同时在其它应用领域，如IP电话等也有参考价值。
AMR声码器除了使用ACELP算法以外，还大量使用了其它控制手段。这些手段包括活动语音探测，舒适噪声，速率切换，多带滤波等等。活动语音探测的概念能够降低语音数据的实际传输速率，舒适噪声能够弥补一定情况下语音自然度下降的问题。这些工程技术与算法构成了声码器的核心。
本文第二章详细介绍了ACELP语音编码的基本原理。第三章主要讨论AMR声码中的各种控制机制。第四章着重讨论了AMR声码器的变速率工作部分。最后给出了对该声码器模型的计算机软件仿真结果及性能分析。
One of main services in mobile communication is speech and the key technology is the coding and decoding of speech signal. Now, the low-rate speech coding has been the main research aspect so as to save frequency bands as many as possible. In recent years, coding algorithm (rate from 4. Skbps to 16kbps) usually use Code Excitation Linear Predictive(CELP) .
With the development of the third generation mobile communication technology, next generation speech coding technology has been set up. Algebra Code Excitation Linear Predictive (ACELP) algorithm which is mature upon rate 4kbps has been chosen as the standard algorithm in 3G Adaptive Multi Rate(AMR) speech coder. To fit the third generation mobile communication, AMR coder can work in several mode with rate range from 4. 75kbps to 12. 2kbps. ACELP is widely used in AMR speech coder. ACELP algorithm has been proved high coding efficiency, low complexity and working stability. Many new generation speech coding standard has chose it as algorithm, such as ITUT G.723, G.729 etc. This speech coder is important in mobile speech communication in future. It is the kernel of the third generation speech communication.
In AMR speech coder, many other techniques have been used beside ACELP algorithm. These techniques include voice activity detection, source controlled rate operation, comfort noise, rate switch, multi-band filtering etc. Voice activity detection can reduce actual transmission rate of speech data, comfort noise can improve natureness of speech. These engineering method and algorithm make up of the coder's core.
The second chapter introduces the basic theory of ACELP speech coder. The third chapter mainly introduce controlling mechanisms.
The forth chapter focus on rate switch in AMR coder. Finally, we give the computer emulation result and performance analyzation for this AMR coder.
随着第三代无线移动通信技术的发展与成熟，下一代语音编码技术标准也基本定型。3G AMR声码器采用目前在4K以上速率中应用比较成熟的ACELP技术，并使用活动语音探测和舒适噪声等控制方式提高编码器的主观听觉性能和频带的利用率。为了适应未来第三代无线通信的需要，AMR可以在多种工作模式之间切换，速率从4.75kbps到12.2kbps共8种，因而能够适用在各种不同的用途中。AMR声码器采用了广泛使用的ACELP技术。经过长期实践验证，ACELP算法具有编码效率高，算法复杂度低和工作稳定等主要特点。许多新一代的语音编码标准方案中都采用了该算法，例如国际电联的G.723，G.729等。该声码器在未来的移动通信话音中占有重要地位，是第三代语音通信的核心，同时在其它应用领域，如IP电话等也有参考价值。
AMR声码器除了使用ACELP算法以外，还大量使用了其它控制手段。这些手段包括活动语音探测，舒适噪声，速率切换，多带滤波等等。活动语音探测的概念能够降低语音数据的实际传输速率，舒适噪声能够弥补一定情况下语音自然度下降的问题。这些工程技术与算法构成了声码器的核心。
本文第二章详细介绍了ACELP语音编码的基本原理。第三章主要讨论AMR声码中的各种控制机制。第四章着重讨论了AMR声码器的变速率工作部分。最后给出了对该声码器模型的计算机软件仿真结果及性能分析。
One of main services in mobile communication is speech and the key technology is the coding and decoding of speech signal. Now, the low-rate speech coding has been the main research aspect so as to save frequency bands as many as possible. In recent years, coding algorithm (rate from 4. Skbps to 16kbps) usually use Code Excitation Linear Predictive(CELP) .
With the development of the third generation mobile communication technology, next generation speech coding technology has been set up. Algebra Code Excitation Linear Predictive (ACELP) algorithm which is mature upon rate 4kbps has been chosen as the standard algorithm in 3G Adaptive Multi Rate(AMR) speech coder. To fit the third generation mobile communication, AMR coder can work in several mode with rate range from 4. 75kbps to 12. 2kbps. ACELP is widely used in AMR speech coder. ACELP algorithm has been proved high coding efficiency, low complexity and working stability. Many new generation speech coding standard has chose it as algorithm, such as ITUT G.723, G.729 etc. This speech coder is important in mobile speech communication in future. It is the kernel of the third generation speech communication.
In AMR speech coder, many other techniques have been used beside ACELP algorithm. These techniques include voice activity detection, source controlled rate operation, comfort noise, rate switch, multi-band filtering etc. Voice activity detection can reduce actual transmission rate of speech data, comfort noise can improve natureness of speech. These engineering method and algorithm make up of the coder's core.
The second chapter introduces the basic theory of ACELP speech coder. The third chapter mainly introduce controlling mechanisms.
The forth chapter focus on rate switch in AMR coder. Finally, we give the computer emulation result and performance analyzation for this AMR coder.
引文
[1] GSM 03． 50 : "Digital cellular telecommunications system (Phase 2) ; Transmission planning aspects of the speech service in the GSM Public Land Mobile Network (PLMN) system".
[2] 3G TS 26． 090 : "AMR Speech Codec; Transcoding functions".
[3] 3G TS 26． 073 : "AMR Speech Codec; ANSI-C code".
[4] 3G TS 26． 074 : "AMR Speech Codec; Test sequences".
[5] 3G TS 26． 093 : "AMR Speech Codec; Source Controlled Rate operation".
[6] 3G TS 26,094 : "AMR Speech Codec; Voice Activity Detection (VAD)".
[7] 3G TS 26． 092 : "AMR Speech Codec; Comfort Noise Aspects".
[8] 3G TS 26． 091 : "AMR Speech Codec; Error Concealment of Lost Frames.
[9] 3G TS 26． 101 : "AMR Speech Codec; Frame Structure".
[10] 3G TS 26． 102 : "AMR Speech Codec; Interface to RAN".
[11] TS 26． 901: "AMR Speech Codec;Performance characterisation".
[12] Multirate Digital Signal Processing Ronald E.Crochiere/Lawrence R.Rabiner
[13] DIGITAL PROCESSING OF SPEECH SIGNALS Lawrence R.Rabiner /Ronald W.Schafer
[14] Thomas E.Tremain "The Government Standard Linear Predivtive Coding Algoeithm:LPC-10",Speech Technology April 1982 pp40-49
[15] 杨行峻，迟惠生：“语音信号数字处理”，北京：电子工业出版社， 1995．
[16] D.W.Griffin and J.S.Lim "Multiband Excitation Vocoder," IEEE Tran.on ASSP,Vol.36,No.8,August 1988
[17] J.C.Hardwick and J.S.Lim "A 4800b/s Improved Multi-Band Excitation Speech coder" Proc.of IEEE Workshop on Speech Coding for Tele .,Canada ,Sept.5-8,1989
[18] Tohkura Y.and Itakura F."Spectral smoothing technique in PARCOR speech analysis-synthesis",IEEE Trans,on ASSP.Vol ASSP-26,No.6,Dece.1978,pp 587-596
[19] Soong F.K.and Juang B.H."Line spectrum pair (LSP) and speech data compression" Proc.IEEE ICASSP,March,1984,pp 1． 10． 1-1． 10． 4
[20] Kabal P.and Ramachandran R.P " The computation of line spectral frequenciex using Chebyshev polynomials " IEEE Tran.on ASSP,Vol.ASSP-34,No 6,Dece.1986,pp1419-1426
[21] H.Chen,W.C.Wong and C.C.Ko "Comparion of Pitch Prediction and adaptation algorithms in forward and backward adaptive CELP systems" IEEE PROCEDINGS-I Vol.140,No.4 Aug.1993 pp240-245
[22] 广书：“数字信号处理-理论、算法与实现”，北京：清华大学出版社，1997
[23] aflamme C.,and Adoul J-P.,Salami R.,Morissette S.and Mabilleau P.: "16 kbps wide-band speech coding technique based on algebraic CELP" ,Proc.IEEE ICASSP,May 1991,pp 13-14
[24] .Laflamma,J-P Adoul,H.Y.Su and S Morisette ,"On reducing computational complexity of codebookl seach in CELP coder through the use of algebraic codes" Proc,ICASSP'90,pp 177-180
[25] Kabal P.and Ramachandran R.P "The computation of line spectral frequenciex using Chebyshev polynomials "IEEE Tran.On ASSP,Vol.ASSP-34,No 6,Dece.1986,pp1419-1426
[2] 3G TS 26． 090 : "AMR Speech Codec; Transcoding functions".
[3] 3G TS 26． 073 : "AMR Speech Codec; ANSI-C code".
[4] 3G TS 26． 074 : "AMR Speech Codec; Test sequences".
[5] 3G TS 26． 093 : "AMR Speech Codec; Source Controlled Rate operation".
[6] 3G TS 26,094 : "AMR Speech Codec; Voice Activity Detection (VAD)".
[7] 3G TS 26． 092 : "AMR Speech Codec; Comfort Noise Aspects".
[8] 3G TS 26． 091 : "AMR Speech Codec; Error Concealment of Lost Frames.
[9] 3G TS 26． 101 : "AMR Speech Codec; Frame Structure".
[10] 3G TS 26． 102 : "AMR Speech Codec; Interface to RAN".
[11] TS 26． 901: "AMR Speech Codec;Performance characterisation".
[12] Multirate Digital Signal Processing Ronald E.Crochiere/Lawrence R.Rabiner
[13] DIGITAL PROCESSING OF SPEECH SIGNALS Lawrence R.Rabiner /Ronald W.Schafer
[14] Thomas E.Tremain "The Government Standard Linear Predivtive Coding Algoeithm:LPC-10",Speech Technology April 1982 pp40-49
[15] 杨行峻，迟惠生：“语音信号数字处理”，北京：电子工业出版社， 1995．
[16] D.W.Griffin and J.S.Lim "Multiband Excitation Vocoder," IEEE Tran.on ASSP,Vol.36,No.8,August 1988
[17] J.C.Hardwick and J.S.Lim "A 4800b/s Improved Multi-Band Excitation Speech coder" Proc.of IEEE Workshop on Speech Coding for Tele .,Canada ,Sept.5-8,1989
[18] Tohkura Y.and Itakura F."Spectral smoothing technique in PARCOR speech analysis-synthesis",IEEE Trans,on ASSP.Vol ASSP-26,No.6,Dece.1978,pp 587-596
[19] Soong F.K.and Juang B.H."Line spectrum pair (LSP) and speech data compression" Proc.IEEE ICASSP,March,1984,pp 1． 10． 1-1． 10． 4
[20] Kabal P.and Ramachandran R.P " The computation of line spectral frequenciex using Chebyshev polynomials " IEEE Tran.on ASSP,Vol.ASSP-34,No 6,Dece.1986,pp1419-1426
[21] H.Chen,W.C.Wong and C.C.Ko "Comparion of Pitch Prediction and adaptation algorithms in forward and backward adaptive CELP systems" IEEE PROCEDINGS-I Vol.140,No.4 Aug.1993 pp240-245
[22] 广书：“数字信号处理-理论、算法与实现”，北京：清华大学出版社，1997
[23] aflamme C.,and Adoul J-P.,Salami R.,Morissette S.and Mabilleau P.: "16 kbps wide-band speech coding technique based on algebraic CELP" ,Proc.IEEE ICASSP,May 1991,pp 13-14
[24] .Laflamma,J-P Adoul,H.Y.Su and S Morisette ,"On reducing computational complexity of codebookl seach in CELP coder through the use of algebraic codes" Proc,ICASSP'90,pp 177-180
[25] Kabal P.and Ramachandran R.P "The computation of line spectral frequenciex using Chebyshev polynomials "IEEE Tran.On ASSP,Vol.ASSP-34,No 6,Dece.1986,pp1419-1426