Showing papers on "Code-excited linear prediction published in 2013"

Residual DPCM for lossless coding in HEVC

Abstract: Incorporating sample-based prediction during lossless coding can significantly improve coding performance. However, its use within a codec designed for lossy coding requires a modification of the available prediction scheme. When implementing the codec, two different prediction processes will have to be implemented. This paper describes a lossless coding scheme that delays the sample-based prediction till the residue coding stage of the codec and carries out prediction in the residual domain. In this way, the prediction scheme of the lossy coder can be retained while realizing the coding gains associated with sample-based prediction. The proposed scheme improves lossless intra coding performance in HEVC Main Profile by an average of 6.5%.

Real-time implementations of sparse linear prediction for speech processing

Abstract: Employing sparsity criteria in linear prediction of speech has been proven successful for several analysis and coding purposes. However, sparse linear prediction comes at the expenses of a much higher computational burden and numerical sensitivity compared to the traditional minimum variance approach. This makes sparse linear prediction difficult to deploy in real-time systems. In this paper, we present a step towards real-time implementation of the sparse linear prediction problem using hand-tailored interior-point methods. Using compiled implementations the sparse linear prediction problems corresponding to a frame size of 20ms can be solved on a standard PC in approximately 2ms and orders faster than with general purpose software.

Scalable Speech Coding for IP Networks: Beyond iLBC

Abstract: High quality speech at low bit rates makes code excited linear prediction (CELP) the dominant choice for a narrowband coding technique despite the susceptibility to packet loss. One of the few techniques which received attention after the introduction of CELP coding technique is the internet low bitrate codec (iLBC) because of inherent high robustness to packet loss. Addition of rate flexibility and scalability makes the iLBC an attractive choice for voice communication over IP networks. In this paper, performance improvement schemes of multi-rate iLBC and its scalable structure are proposed, and the proposed codec enhanced from the previous work is re-designed based on the subjective listening quality instead of the objective quality. In particular, perceptual weighting and the modified discrete cosine transform (MDCT) with short overlap in weighted signal domain are employed along with the improved packet loss concealment (PLC) algorithm. The subjective evaluation results show that the speech quality of the proposed codec is equivalent to that of state-of-the-art codec, G.718, under both a clean channel condition and lossy channel conditions. This result is significant considering that development of the proposed codec is still in early stage.

Sparse signal modeling in a scalable CELP coder

Abstract: This work presents a series of sparse signal modeling algorithms implemented in a variable rate CELP coder in order to compare their performances at a reasonable computational load. Multipulse excitation (MPE), Multi-Pulse Maximum Likelihood Quantization (MP-MLQ), Algebraic CELP (ACELP) and hybrid excitation schemes are analyzed under a common framework. New approaches are proposed, based on cyclic and parallel use of fast greedy algorithms. These algorithms yield a statistically significant reduction of signal approximation error at a controllable computational complexity. Main results were confirmed by comparing MOS values obtained with the PESQ algorithm

Improved speaker identification in wireless environment

Abstract: The increasing use of wireless mobile systems is creating great deal of interest in the development of robust speech systems in wireless environment. The major degradations involved in wireless environment are: acoustic environment, speech coding and transmission errors. In this paper, we address the problem of Speaker Identification (SI) from coded and cellular speech. Since speaker-specific characteristics are preserved in steady vowel segments of speech even after coding, the features extracted from these steady vowel regions are used to build the SI system. We have proposed a method to determine the steady vowel region from the speech signal by using vowel onset points and epochs. SI studies are carried out using cellular, coded and microphone speech databases. Autoassociative Neural Network (AANN) models are explored for developing the SI models. Speech coders considered in this work are GSM CELP, and MELP. Significant improvement in the performance of SI system is observed by using proposed approach.

Packet loss concealment by interpolation for speech over IP network services

Abstract: This paper presents a packet loss concealment (PLC) method based on interpolation by separation odd and even frames to improve speech quality deterioration caused by packet losses for CELP based coders. We applied our scheme to the standard ITU-T G.729 speech coder to evaluate the proposed method. The perceptual evaluation of speech quality (PESQ) and enhanced modified bark spectral distortion (EMBSD) tests under various packet loss conditions confirm that the proposed algorithm is superior to the concealment algorithm embedded in the G.729. The performance measures prove that our concealment method based interpolation is better. (4 pages)

Low-complexity PARCOR coefficient quantization and prediction order estimation designed for entropy coding of prediction residuals

Abstract: This paper presents a set of low-complexity tools used in lossless coding of G.711 bitstream, based on linear prediction. One is an algorithm for quantizing the PARCOR/reflection coefficients and the other is an estimation method for the optimal prediction order. Both tools are based on a criterion that minimizes the entropy of the prediction residual signal and can be implemented in fixed-point arithmetic at very low-complexity. Since proposed methods show efficient performance in terms of compression and complexity, they are adopted in the Recommendation ITU-T G.711.0, a new standard for lossless compression of G.711 (A-law/ -law logarithmic PCM) payload.

Efficient speech encryption using chaotic cat map for code-excited linear prediction based coders in packet networks

Abstract: The increasing importance of multimedia applications is placing a great insist on content protection and customer privacy. Communications can be intercepted, especially over wireless links. Since encryption can effectively prevent eavesdropping, its use is widely advocated. The codec G. 729 based CS-ACELP algorithm is standardized as voice codec by ITU-T for multimedia and Voice over Internet Protocol (VoIP) applications. In this paper, we introduce a speech encryption method based chaotic cat map algorithm. Cat map extended to two-dimensional NxN matrix. It takes concepts from linear algebra and uses them to change the positions of the values of the matrix. The result after applying the Cat Map will be shuffled signals that contain the same values of the original signals. We applied our encryption scheme to the standard ITU-T G.729 standard speech coder to evaluate its performance. Simulation results show that G.729 based cat map encryption is very efficient since the encrypted speech is similar to a white noise. The perceptual evaluation of speech quality (PESQ) and enhanced modified bark spectral distortion (EMBSD) tests for speech extracted from TIMIT database confirm the efficiency of our proposed scheme.

A new speech coding algorithm using zero cross and phoneme based SYMPES

Abstract: In this work, a new low bit rate hybrid speech coding approach which combines the benefits of the SYMPES (Systematic Procedure for Predefined Envelope and Signature Sequences) and zero cross and phoneme based segmentation is proposed. In the new approach, the SYMPES structure is developed in the phoneme based fashion. In order to achieve lower bit rates, some drawbacks such as computational complexity, relatively high encoding times etc. of the SYMPES are also eliminated in the new version. Experimental results show that in almost same bit rates very promising speech quality is obtained compared to the other conventional methods such as CELP (Code Excited Linear Predictive) coding algorithm.

Development of high quality speech compression system for Quranic recitation based on modified CELP algorithm

Abstract: Currently, there are more than 500 Quranic recitations available freely on the internet. There is also a growing trends on the use of smart phone compare to traditional desktop PCs for accessing the internet. On such limited device, a high quality speech compression for Quranic recitation is favorable. In this paper, we developed a high quality speech compression for Quranic recitation by modifying Code Excited Linear Prediction (CELP) algorithm. First, the characteristics of the Quranic recitation of all surahs was evaluated in which it was found that the voiced speech is more dominant compare to unvoiced speech. Next, we optimize CELP algorithm based on previous findings by modifying the original stochastic codebook. Instead of random Gaussian signal, we trained the codebook using LBG algorithm for Surah Al-Fatihah, and used the trained codebook for the whole Quran. Lastly, we evaluate the performance of the developed algorithm objectively using PESQ (Perceptual Evaluation of Speech Quality). Results showed that our proposed algorithm performs better than the traditional CELP algorithm, in terms of PESQ score and processing time.

A 600bps MELP-Based Speech Quantization Scheme for Underwater Acoustic Channels

Abstract: The MELP is the U.S. Federal Standard operated at 2.4kbps, it could achieve equal to or better than the performance of the 4.8kbps CELP vocoder. However, for channels like underwater acoustic channel, the condition is very bad that even the 2.4kbps MELP data could not be transferred. A new 600 bps quantization method based on the MELP parameters is proposed. The new scheme contains methods like inter-frame prediction, linear interpolation and so on. Though the bit rate is only a quarter of the original MELP algorithm, the scheme could also synthesize a reconstruction speech which is understandable.

A new bandwidth extension technology for MPEG Unified Speech and Audio Coding

Abstract: In January 2012, MPEG finalized the new MPEG-D Unified Speech and Audio Coding (USAC) standard, which enables the coding of a variety of audio content at low bitrates. USAC provides low-bitrate coding by integrating a speech codec and an audio codec into a unified system. In USAC, Predictive Vector Coding (PVC) is added to Enhanced Spectral Band Replication (eSBR) to improve the subjective quality, especially for speech at low bitrates. For speech signals, there is generally a relatively high correlation between the spectral envelopes of low- and high-frequency bands. The PVC scheme exploits this by predicting the high-frequency envelopes from the low-frequency ones, with the coefficient matrices for the prediction being coded by means of vector quantization.

Analysis of detection of vowel offset point for coded speech

Abstract: Vowel onset and vowel offset points are the instants at which the onset and offset of vowel take place in the speech signal. Vowel regions start with the vowel onset point and end with the vowel offset point. This paper discusses the effect of speech coding on detection of vowel offset point. Speech coding is the one of the major degradation in mobile environment. In this work, effect of speech coding is studied by using two recently developed methods, COMB-ESM and SE-GC for vowel offset point detection. COMB-ESM method uses the combination of evidences from excitation source, spectral peaks energy and modulation spectrum. SE-GC method uses spectral energy within glottal closure region for detecting the vowel offset points. Speech coders used in this study are GSM full rate (ETSI 06.10), CELP (FS-1016), and MELP (TI 2.4 kbps). Performance of vowel offset point detection methods is evaluated using TIMIT database and consonant-vowel (CV) units collected from the broadcast news corpus. Experimental results show that performance of the SE-GC method is better compared to COMB-ESM method under speech coding.

Backwards-compatible error propagation recovery for the amr codec over erasure channels

Abstract: This paper presents a recovery scheme for the error-propagation distortion which frequently appears after a frame erasure in CELP-based speech coders, in particular the AMR codec. The extensive use of predictive filters and parameter encoding allow a high-quality speech synthesis in these codecs, but makes them more vulnerable to frame erasures. Thus, when a frame is lost, an additional distortion appears in the subsequent frame, although that was correctly received, further degrading the speech quality. This degradation can also propagate over several frames, being even more damaging than the loss itself. This well known fact has motivated the development of techniques which prevent or mitigate the error propagation. Nevertheless, the previously proposed methods in some respect modify the transmission scheme (by including additional frames, FEC codes, etc.) making them incompatible with the original decoder. In this work, we apply a steganographic technique to embed recovery data to assist the decoder after a frame loss. This data mainly consist of resynchronization pulses and correction vectors for the excitation signal and the spectral envelope, respectively. PESQ results confirm that our proposal achieves a higher robustness against error propagation while the full backwards-compatibility with the AMR standard is retained.

Development of Low Bit Rate Speech Encoder based on Vector Quantization and Compressive Sensing

Abstract: Speech coding is a representation of a digitized speech signal using as few bits as possible, while maintaining reasonable level of speech quality. Due to growing need for bandwidth conservation in wireless communication, the research in speech coding has increased. Recently, Compressive Sensing (CS) is gaining a great interest because of its ability to recover original signals by taking only few measurements. CS is a new approach that goes against the common data acquisition methods. In this research, a new system of speech encoding system is developed using compressive sensing. Since CS performs well in sparse signals, different sparsifying transforms are analyzed and compared using Gini coefficient. The quality of the speech coder is evaluated using Perceptual Evaluation of Speech Quality (PESQ), Signal-to-Noise Ratio (SNR) and subjective listening tests. Results show that the speech coders have achieved a PESQ score of 3.16 at 4 kbps which is a good quality as confirmed by listening tests. Furthermore, the coder is also compared with Code Excited Linear Prediction (CELP) coder.

Packet loss concealment using piggybacking for speech over IP network services

Abstract: This paper addresses a packet loss concealment method (PLC) based on piggybacking to improve speech quality degradation caused by packet losses for code excited linear predictive (CELP) type coders. We applied our proposed scheme to the standard ITU-T G.729 Conjugate-Structure Algebraic CELP (CS-CELP) speech coder to evaluate its performance. The average spectral distortion (Avg. SD), the perceptual evaluation of speech quality (PESQ) and enhanced modified bark spectral distortion (EMBSD) tests under a variety of packet loss rates prove that the proposed PLC based piggybacking is better than the concealment method embedded in the ITU-T G.729 for speakers (female and male). The performance measures confirm that our PLC method based piggybacking is superior than the one embedded in the standard ITU-T G.729.

State-space approach to linear predictive coding of speech — A comparative assessment

Abstract: Speech coders are fundamental component in telecommunication and multimedia infrastructure. Several systems like, mobile telephony, voice over internet protocol (VOIP), audio conferencing etc., rely on efficient speech coding. Speech coders strive to provide low-bit rate maintaining the same speech quality and intelligibility. Linear predictive coding uses spectral properties of the speech to “optimize” the coder's performance for human ear. In this paper we perform a comparative assessment of speech coding performance of some state-space filters to give designers an insight into capabilities of these filters. The filters considered are Kalman filter, state-space recursive least-squares (SSRLS) and SSRLS with adaptive memory (SSRLSWAM). The results of RLS and LMS are also quoted. The performance is judged in terms of perceptual evaluation of speech quality (PESQ) and prediction gain.

Exploring the gap between a pre- and post-installation of a corporate e-learning program in an accounting workplace

Abstract: Although earlier concentration has addressed the use of corporate e-learning programs (CELP), the dissimilitude between pre and post installation reaction to CELP is less explored. This study adopted a two-phase investigation to survey learner intention to use CELP and actual behavior within an international accounting firm. In the pre-installation phase, a survey was conducted to evaluate learner intention to use CELP, followed one year later by a second investigation to examine learner reactions to CELP and the actual usage frequency/ duration. The results of this study identified there is actually a difference between intention and actual usage duration. Further questionnaire surveys were implemented to identify learner reactions and factors that could potentially contribute to the gap between intention and actual usage duration. Results also indicated that scheduling was the critical factor leading to the differences in actual usage. The conclusions clarify the relationships among learner intention to use CELP, actual usage frequency/ duration, and subsequent reactions towards it.

Design of nonlinear predictors for adaptive predictive coding of speech signals

Abstract: Linear predictive coding is probably the most frequently used technique in speech signal processing. Its main advantage comes from the analogy of the simplified vocal tract model with speech production system. However, this neglects nonlinearities in the speech production process. The paper deals with nonlinear prediction of speech based on truncated Volterra series. Long-term one-tap Volterra predictor is designed in order to decrease computational complexity. Further improvements are obtained using frame/subframe structure and fractional delay.

Joint linear prediction and epoch estimation of voiced speech using a basis where the prediction residual can be sparsely represented

Abstract: Whereas most approaches to linear speech prediction fail to account for the quasi-periodic glottal flow, this paper incorporates the Liljencrants-Fant model for glottal flow derivative (GFD) directly into the linear prediction problem. A linear model for the prediction error is obtained by constructing a dictionary of time-shifted GFD pulses. Minimizing the difference between the linear prediction residual and a sparse combination of the pulses in the dictionary leads to joint estimation of the linear predictor as well as a sparse representation for the prediction error that reveals the instants of vocal tract excitation (epochs). A greedy algorithm is proposed to approximately solve this joint estimation problem. The method is applied to voiced segments extracted from the CMU Arctic dataset which also includes electro-glottograms. Results show that the approach and the proposed algorithm are effective in estimating the parameters of interest.

The effect of the spoken language on the distortion of the linear prediction coders

Abstract: The newly developed Code Excited Linear Prediction "CELP" Speech coders combine low data rates and good speech quality. However these coders have been built initially for 7 languages not included Arabic language. So what is the effect of the change of the spoken language or accent? This paper answers on many questions. The first question is; what is the effect of the language or accents on CELP coders? Moreover what will happen if the speech is compressed more by lower data rate coder and at the same time the language is other than English? Finally; what is the defective part in the coder? Extensive testing is done on 3 coders ITU G.711, ITU G.723.1 and 3GPP AMR. The outputs were compared with the ITU PESQ algorithm. It has been proved that the speech quality will be slightly degraded when using Arabic. Also the quality is decreased as the compression ratio increased which will be lower and unstable for Arabic or Cairo accent. Finally it is found that the main problem is the Linear Prediction vector quantization CodeBook and has been verified by the MFCC and LBG algorithm.

Speech signal compression analyses based prediction and coding

Abstract: In this paper, the main analysis prospective investigated is to compute the effect of the Digital compression of speech signal based on linear prediction and coding can increase the speed of data transmission using coding one type of such coding is Differential pulse-code modulation. In the latter case. Approach: Linear model provides a reliable predictor of a reliable and accurate method of estimating the parameters that characterize the linear, time-varying system. In this project, we are implementing a voice excited vocoder for low compression bit speech

Nonlinear bi-directional prediction model of speech signals based on genetic programming

Abstract: Based on the linear prediction theory, This paper proposes a nonlinear bi-directional prediction model for speech signals. After the data preprocessing, an improved GP is used to construct bi-directional prediction model of each frame. Then by the analysis of these models, the normalized nonlinear bi-directional prediction model is obtained. In the experiments, the DUPSO algorithm is used to optimize the parameters. By comparing the SNR and MMSE with the linear prediction model, the superiority of the proposed nonlinear bi-directional prediction model is proved.

Channel prediction parameter selection for multi-channel audio coding

Abstract: An audio coding device that uses a first-channel signal, a second-channel signal, and a plurality of channel prediction coefficients included in a code book, according to which predictive coding is performed on a third-channel signal, the first-channel signal, the second-channel signal, and the third-channel signal being included in a plurality of channels of an audio signal, the device includes, a determining unit that determines a distribution of error defined by a difference between the third-channel signal before predictive coding and the third-channel signal after predictive coding as a given curved surface according to the first-channel signal, the second-channel signal, and the third-channel signal before predictive coding; and a calculating unit that calculates channel prediction coefficients, included in the code book, that correspond to the first channel and the second channel from the code book.

Employing Linear Prediction Coding in Feature Time Sequences for Robust Speech Recognition in Noisy Environments.

Abstract: 近幾十年來,無數的學者先進對於此雜訊干擾問題提出了豐富眾多的演算法,略分成兩 大類別:強健性語音特徵參數表示法(robust speech feature representation)與語音模型調適 法(speech model adaptation),第一類別之方法主要目的在抽取不易受到外在環境干擾下 而失真的語音特徵參數,或從原始語音特徵中儘量削減雜訊造成的效應,比較知名的方 法有:倒頻譜平均值與變異數正規化法 (cepstral mean and variance normalization, CMVN)[1]、倒頻譜統計圖正規化法(cepstral histogram normalization, CHN)[2]、倒頻譜平 均值與變異數正規化結合自動回歸動態平均濾波器法(cepstral mean and variance normalization plus auto-regressive-moving average filtering, MVA)[3]等;第二類別之方 法,則藉由少量的應用環境語料或雜訊,來對原始的語音模型中的統計參數作調整,降 低模型之訓練環境與應用環境之不匹配的情況。較有名的語音模型調適技術包含了:最 大後機率法則調適法(maximum a posteriori adaptation, MAP)[4]、平行模型合併法(parallel model combination, PMC)[5]、向量泰勒級數轉換(vector Taylor series transform, VTS)[6] 等。本論文較集中討論與發展的是上述的第一類方法,我們提出一套作用於倒頻譜時間 序列域的強健性技術,稱作線性估測編碼濾波法(linear prediction coding-based filtering, LPCF),此方法主要是應用線性估測(linear prediction)[7]的原理,來擷取語音特徵隨著 時間變化的特性、進而凸顯語音的成分、抑制雜訊的成分。在 LPCF法中,將一段時域 (time domain)上的訊號 [ ] x n 用以下數學式表示:

Artificial Bandwidth Extension Based Packet Loss Concealment for CELP-Type Speech Coders

Abstract: In this paper, a packet loss concealment (PLC) algorithm is proposed to improve the quality of decoded speech when packet losses occur in a code- excited linear prediction (CELP)-type speech coder. The proposed PLC algorithm is based on artificial bandwidth extension (ABE) from narrowband to wideband, consisting of packet loss concealment in the narrow-band, ABE in the modified discrete cosine transform (MDCT) domain, and smoothing of wideband MDCT coefficients with those of the last good frame. The effectiveness of the proposed PLC algorithm is demonstrated by an informal listening test.