Preface. Acronyms. Notation. Introduction. Signal Processing Techniques. Stochastic Processes and Models. Linear Prediction. Scalar Quantization. Pulse Code Modulation and Its Variants. Vector Quantization. Scalar Quantization of Linear Prediction Coefficient. Linear Prediction Coding. Regular-Pulse Excitation Coders. Code-Excited Linear Prediction. The Federal Standard Version of CELP. Vector Sum Excited Linear Prediction. Low-Delay CELP. Vector Quantization of Linear Prediction Coefficient. Algebraic CELP. Mixed Excitation Linear Prediction. Source-Controlled Variable Bit-Rate CELP. Speech Quality Assessment. Appendix A: Minimum-Phase Property of the Forward Prediction-Error Filter. Appendix B: Some Properties of Line Spectral Frequency. Appendix C: Research Directions in Speech Coding. Appendix D: Linear Combiner for Pattern Classification. Appendix E: CELP: Optimal Long-Term Predictor to Minimize the Weighted Difference. Appendix F: Review of Linear Algebra: Orthogonality, Basis, Linear Independence, and the Gram-Schmidt Algorithm. Bibliography. Index.

Speech coding algorithms : foundation and evolution of standardized coders

The program Praat is a versatile speech analysis program, which becomes a standard in speech/language analysis scientific community. Automatic speech segmentation has been developed and used in Praat using a plugin called EasyAlign, which requires and input text (separated sentences) and clean audio recordings with quiet pauses between the sentences to operate. In this paper, a new functions will be evaluated and implemented into Praat to improve automatic speech segmentation processes and enables to process also noisy audio recordings with no input texts available. One of them is using also our own cloud based ASR (Automatic Speech recognition) solution developed. The results of the automatic speech recognition and different audio processing (noise-gate, de-noiser, normalization) were added into the Praat system to expand the EasyAlign functionality and improve the resulted automatic speech segmentation. The graphical output of the Praat also helps to visualize the automatic speech segmentation and also automatic speech recognition results for educational and system developers subjective evaluation purposes.

Automatic Acoustic Speech segmentation in Praat using cloud based ASR

Endangered wildlife is protected in remote land where people are restricted to enter. But intrusions of poachers and illegal loggers still occur due to lack of surveillance to cover a huge amount of land. The current usage of stealth ability of the camera is low due to limitations of camera angle of view. Maintenance such as changing batteries and memory cards were troublesome reported by Wildlife Conservation Society, Malaysia. Remote location with no cellular network access would be difficult to transmit video data. Rangers need a system to react to intrusion on time. This paper aims to address the development of an audio events recognition for intrusion detection based on the vehicle engine, wildlife environmental noise and chainsaw activities. Random Forest classification and feature extraction of Linear Predictive Coding were employed. Training and testing data sets used were obtained from Wildlife Conservation Society Malaysia. The findings demonstrate that the accuracy rates achieve up to 86% for indicating an intrusion via audio recognition. It is a good attempt as a primary study for the classification of a real data set of intruders. This intrusion detection will be beneficial for wildlife protection agencies in maintaining security as it is less power consuming than the current camera trapping surveillance technique.

Acoustic Surveillance Intrusion Detection with Linear Predictive Coding and Random Forest

Effective audio communication is essential in supporting space flight programs. Information exchange between space and ground links plays a vital role to ensure a successful space mission. However, the availability of bandwidth for data transmission is mission specific. A compression technique, which achieves data compression by removing data redundancy, is used to reduce transmission bandwidth required for huge data transfer. On comparing various Consultative Committees for Space Data Systems (CCSDS) audio compression standards, it has been inferred that the speech codec G.729E manages to achieve “toll” quality speech. The bit rate for this codec is as low as 8 Kbps. It delivers audio signals in small packets of only 10ms time duration. A compression ratio of 16:1 can be achieved using this codec. The algorithm used in G.729E is Conjugate-Structure Algebraic Code Excited Linear Prediction (CS-ACELP) for extracting speech-characterizing parameters. Implementation of this highly complex algorithm on an FPGA would reduce processing delays through parallel processing and pipelining. In this paper linear prediction coefficient in G.729E encoder is implemented on a space graded FPGA RTAX2000S.

Implementation of Linear Prediction Coefficients in G.729E Using VHDL for Man Mission Applications

Linear predictive coders form an important class of speech coders. This paper describes the software level implementation of linear prediction based vocoders, viz. Code Excited Linear Prediction (CELP), Low-Delay CELP (LD-CELP) and Mixed Excitation Linear Prediction (MELP) at bit rates of 4.8 kb/s, 16 kb/s and 2.4 kb/s respectively. The C programs of the vocoders have been compiled and executed in Linux platform. Subjective testing with the help of Mean Opinion Score test has been performed. Waveform analysis has been done using Praat and Adobe Audition software. The results show that MELP and CELP produce comparable quality while the quality of LD-CELP coder is much higher, at the expense of higher bit rate.

Performance Comparison of Linear Prediction based Vocoders in Linux Platform

This paper describes the analysis of the audio and speech output of the 16 kb/s Low Delay CELP algorithm (LD-CELP). For various speech file conversions, the SoX (Sound eXchange) tool is used. Praat software is used to extract details regarding the quality of the sound synthesized by the coder. The C program of the coder was compiled using the GCC compiler in the Linux environment and executed using the GNU Make utility. File compression ratio of 5 was achieved. The bit rate obtained is 16 kb/s, i.e. a reduction by 4 times of the original bitrate. The perceptual quality of the reconstructed audio output is very good, and this is evident from the spectrogram, pitch, intensity and formant waveforms obtained.

Analysis of LD-CELP coder output with Sound eXchange and Praat software

Attractive improvements have been made during these days in coding speech with high quality at low bit rates and low delay. The need for low bit rate speech coding algorithms continues, supported by the ever increasing number of users to the wireless communication networks. This paper discusses the implementation details of Code Excited Linear Prediction Speech Coder at different bit rates (16Kbps, 9.6Kbps, 7Kbps, 6.8Kbps, 4.9Kbps and 4.8Kbps) and analytical evaluation of performance in terms of bit rate and quality using PRAAT software.

Performance evaluation of Code Excited Linear Prediction Speech Coders at various bit rates

The design of high quality speech coders with low data rate is a very challenging task in speech processing. CELP provides good quality coded speech at a very low bit rate of 4.8kbps. Two types of codebooks namely Gaussian codebook and Fixed codebook are used in this analysis. The changes in the performance of the CELP codec is evaluated for theses two codebooks. Optimization of the codebook can be done by improving the training process in the codebook generation stage. Two different types of codebooks, the Gaussian codebook and the fixed codebook is generated to be implemented in the G723.1 CELP codec and the variation in the performance is evaluated.

Performance analysis of CELP codec for Gaussian and fixed codebooks

The development of high quality speech coders at low bit rate and low delay is a challenging task in the field of speech communication, due to ever increasing users in wireless communication networks. CELP codec gives good quality at 4.8 kbps. By optimizing the excitation vectors in the codebook many changes in the performance can be observed to achieve toll quality speech. Codebook optimization can be achieved by improving the training process in the generation process and by using fast searching algorithms. Gaussian (X-means clustering), Fixed (PSVQ) and PSVQ-Genetic codebooks were generated to be included in CELP 4.8. Variations in performance were evaluated using objective and subjective quality measures. Experiments show that CELP 4.8 provides toll quality and reduced delay with PSVQ-Genetic codebook.

Ancy S. Anselam

Papers

Analysis of LD-CELP coder output with Sound eXchange and Praat software

Performance evaluation of Code Excited Linear Prediction Speech Coders at various bit rates

Performance analysis of CELP codec for Gaussian and fixed codebooks

Performance Comparison of Linear Prediction based Vocoders in Linux Platform

Optimization of code excited linear prediction speech coder with PSVQ-genetic codebook