The implementation and optimization of AMR speech codec on DSP

This letter proposes an efficient uplink scheduling algorithm for voice over Internet protocol (VoIP) services with adaptive multi-rate (AMR) speech codec in IEEE 802.16e/m systems. The proposed scheduling algorithm adopts the random access scheme during silent-period to reduce the waste of uplink bandwidth considering the characteristics of AMR speech codec. The numerical results show that the proposed algorithm can increase the maximum supportable number of voice users by 26% compared to the conventional extended real-time polling service (ertPS).

In today's wireless communication system, quality of decoded speech at receiving end is found muffled and thin, mainly attributed to inherent band limitation (300-3400 Hz) and power constraints. In order to obtain toll quality of recovered speech in terms of intelligibility and naturalness in wireless systems, Narrowband (NB) speech coders should be upgraded to its counterpart Wideband (WB) coders (50-7000Hz). In the meantime, a novel and backward compatible solution is proposed that claims to artificially extend bandwidth of NB speech to WB at receiving end, popular as Artificial Bandwidth Extension (ABE). Out of many techniques which aim to mitigate the effect of the ever unpredictable channel conditions, Adaptive Multi Rate (AMR) NB coder is considered to be one of the potential candidates. Selection of particular bit rate mode (out of all eight bit rate modes between 4.75 kbps and 12.2 kbps) solely depends upon the channel condition. This paper discusses development of ABE algorithm for Code Excited Linear Prediction (CELP) based GSM AMR NB coder, and for the same, MATLAB based e-test bench is created for simulation. Such series of simulations are conducted to discover and gain insight about the overall performance of proposed ABE coder that includes subjective (Mean Opinion Score - MOS) and objective (Perceptual Evaluation of Speech Quality - PESQ) analyses. The evaluated results for both analyses clearly advocate that proposed ABE coder outperforms legacy GSM AMR 06.90 NB coder.

The paper presents a speech generative model that provides an efficient way of generating speech waveform from its amplitude spectral envelopes. The model is based on hybrid speech representation that includes deterministic (harmonic) and stochastic (noise) components. The main idea behind the approach originates from the fact that speech signal has a determined spectral structure that is statistically bound with deterministic/stochastic energy distribution in the spectrum. The performance of the model is evaluated using an experimental low-bitrate wide-band speech coder. The quality of reconstructed speech is evaluated using objective and subjective methods. Two objective quality characteristics were calculated: Modified Bark Spectral Distortion (MBSD) and Perceptual Evaluation of Speech Quality (PESQ). Narrow-band and wide-band versions of the proposed solution were compared with MELP (Mixed Excitation Linear Prediction) speech coder and AMR (Adaptive Multi-Rate) speech coder, respectively. The speech base of two female and two male speakers were used for testing. The performed tests show that overall performance of the proposed approach is speaker-dependent and it is better for male voices. Supposedly, this difference indicates the influence of pitch highness on separation accuracy. In that way, using the proposed approach in experimental speech compression system provides decent MBSD values and comparable PESQ values with AMR speech coder at 6,6 kbit/s. Additional subjective listening testsdemonstrate that the implemented coding system retains phonetic content and speaker’s identity. It proves consistency of the proposed approach.

In this paper we present a voicing controlled, speech parameter based frame loss concealment for frames that have been encoded with the Adaptive Multi-Rate (AMR) speech codec. The missing parameters are estimated by interpolation and extrapolation techniques that are chosen in dependence of the voicing state of the speech frames preceding and following the lost frames. The voicing controlled concealment outperforms the conventional extrapolation/muting based approach and it shows a consistent improvement over interpolation techniques that do not distinguish between voiced and unvoiced speech. The quality can be further improved if additional information about the predictor states of predictively encoded parameters is available from a redundant transmission in future packets.

This paper describes the VAD (voice activity detection) for controlling DTX (discontinuous transmission) of the GSM AMR (adaptive multi-rate) speech codec. The algorithm is based on spectral estimation and periodicity detection. The VAD contains a 9-band IIR filter bank, which divides input signals into frequency bands. The signal level at each band is calculated. Background noise is estimated in each sub-band. The VAD decision is computed by comparing input signal level and background noise estimate. The algorithm incorporates novel methods to estimate background noise and to detect periodic components based on open-loop pitch gain. A new method is also derived to detect correlated complex signals like music.

This paper presents a coded domain level control technique for the Adaptive Multi-Rate (AMR) speech codec. Level control in the coded domain is done by directly modifying quantized speech parameters. The advantage of the method is an accurate speech level control with minimized system complexity and end-to-end delay when tandem coding can be avoided completely. Requantization optimization functions are derived for scalar and vector quantization of relevant parameters. The experimental results show that the presented technique makes possible to apply a desired gain in the coded domain maintaining high speech quality.

Identification of AMR decompressed audio

In this paper we investigate embedded coding of speech in the CELP (code-excited-linear-prediction) framework. Compared to other known approaches of variable bit rate speech coding, such as the adaptive multi-rate (AMR) codec, embedded coding systems allow bit rate reductions at any point along the communication network, without changes of the encoder and the decoder. Thus, the quality of the decoded speech increases with the amount of received bits. Aiming at a coding scheme, that produces such a hierarchically-structured bit stream, we focus on a decomposition of the excitation signal by means of pyramid coding. To achieve reasonable speech compression, analysis-by-synthesis (AbS) based quantization of the pyramid layers is designed in a CELP-type fashion, called P-CELP (pyramid CELP). Besides an efficient design of fixed algebraic codebooks in each pyramid layer, special attention is payed to the integration of an adaptive codebook. To achieve maximum performance, the proposed coding scheme is used for wideband (50 Hz-7 kHz) speech.

Steganalysis of adaptive multi-rate (AMR) speech is a hot topic for controlling cybercrimes grounded in steganography in related speech streams. In this paper, we first present a novel AMR steganalysis model, which utilizes extreme gradient boosting (XGBoost) as the classifier, instead of support vector machines (SVM) adopted in the previous schemes. Compared with the SVM-based model, this new model can facilitate the excavation of potential information from the high-dimensional features and can avoid overfitting. Moreover, to further strengthen the preceding features based on the statistical characteristics of pulse pairs, we present the convergence feature based on the Markov chain to reflect the global characterization of pulse pairs, which is essentially the final state of the Markov transition matrix. Combining the convergence feature with the preceding features, we propose an XGBoost-based steganalysis scheme for AMR speech streams. Finally, we conducted a series of experiments to assess our presented scheme and compared it with previous schemes. The experimental results demonstrate that the proposed scheme is feasible, and can provide better performance in terms of detecting the existing steganography methods based on AMR speech streams.

Lower rate codes can be constructed from a high rate code by inserting known dummy bits into the information bit sequence before encoding. These so-called insertion convolutional (IC) codes can individually adjust the amount of protection given to the information bits, and can be decoded using the same decoder as for the mother codes. We employed the recursive systematic insertion convolutional (RSIC) codes in adaptive multirate (AMR) speech transmission over GERAN (GSM/EDGE radio access network) 8PSK half rate (HR) voice carriers and good results were obtained. The unequal error protection (UEP) necessary for differently important speech encoded bits is realized and optimized by employing different rates of channel codes, followed by a systematic mapping of code bits into strong and weak bits of 8PSK symbols. As a result, the scheme for AMR speech transmission over 8PSK HR voice carriers significantly outperforms the scheme over GMSK HR voice carriers.

Techniques for improving the performance of CELP (code excited linear prediction)-type speech coders while maintaining reasonable computational complexity are explored. A harmonic noise weighting function, which enhances the perceptual quality of the processed speech, is introduced. The combination of harmonic noise weighting and subsample pitch lag resolution significantly improves the coder performance for voiced speech. Strategies for reducing the speech coder's data rate, while maintaining speech quality, are presented. These include a method for efficient encoding of the long-term predictor lags, utilization of multiple gain vector quantizers, and a multimode definition of the speech coder frame. A 5.9-kb/s VSELP speech coder that incorporates these features is described. Complexity reduction techniques which allow the coder to be implemented using a single fixed-point DSP (digital signal processor) are discussed.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Transform-domain and DSP based secure speech communication system

Modern mobile communications require optimum bandwidth utilization with minimum loss, delay and good quality of speech transmission. This triggered the usage of low bit rate voice codecs, among which CELP codecs inherit the merits of both waveform and source codecs, like toll quality, low bit rate etc. In this paper several hardware and software optimization techniques are presented for efficient implementation of ITU G.729 standard (CS-ACELP, conjugate structure algebraic code excited linear prediction) of 8 Kbit/s bit rate on a real time digital signal processor (DSP), with the aim of overcoming the limitation of computational burden and also scaling this application for enhanced speed to process more channels. These techniques are in general applicable to any speech codec and DSP processor platform.

This paper describes the implementation of the scaleable channel coding scheme (Annex C) for ITU-T G.723.1 speech coder. Speech parameter bits from the encoded frame are reordered and allocated to bit classes of differing sensitivity. Each class is then coded by a rate compatible punctured convolutional code. A rate control algorithm provides scaleability by determining the number of bits and the puncture code per class according to the available channel capacity and speech frame type. At the receiver the bitstream is decoded and errors corrected by a Viterbi decoder and the recovered bitstream reordered back into the original speech frame parameters. Flags, which are generated depending on the error conditions, are passed to the speech decoder to enhance error concealment thereby improving the perceptual performance for the listener. A real-time DSP implementation of this scaleable channel coder is presented. The system is implemented on a TMS320C5X fixed-point DSP chip and integrated together with the speech codec. Efficient architectures are used for implementing the rate control algorithm, convolutional encoder and Viterbi decoder.

As adaptive multi-rate (AMR) speech applications become increasingly widespread, AMR-based steganography presents growing security risks. Conventional steganalysis methods often assume known embedding rates, limiting their practicality in real-world scenarios where embedding rates are unknown. To overcome this limitation, we introduce a novel framework that integrates a multi-scale transformer architecture with multi-task learning for joint classification and regression. The classification task effectively distinguishes between cover and stego samples, while the regression task enhances feature representation by predicting continuous embedding values, providing deeper insights into embedding behaviors. This joint optimization strategy improves model adaptability to diverse embedding conditions and captures the underlying relationships between discrete embedding classes and their continuous distributions. The experimental results demonstrate that our approach achieves higher accuracy and robustness than existing steganalysis methods across varying embedding rates.