Venkatraman S. Atti

Overview of the EVS codec architecture

The recently standardized 3GPP codec for Enhanced Voice Services (EVS) offers new features and improvements for low-delay real-time communication systems. Based on a novel, switched low-delay speech/audio codec, the EVS codec contains various tools for better compression efficiency and higher quality for clean/noisy speech, mixed content and music, including support for wideband, super-wideband and full-band content. The EVS codec operates in a broad range of bitrates, is highly robust against packet loss and provides an AMR-WB interoperable mode for compatibility with existing systems. This paper gives an overview of the underlying architecture as well as the novel technologies in the EVS codec and presents listening test results showing the performance of the new codec in terms of compression and speech/audio quality.

Improved error resilience for volte and VoIP with 3GPP EVS channel aware coding

A highly error resilient mode of the newly standardized 3GPP EVS speech codec is described. Compared to the AMR-WB codec and other conversational codecs, the EVS channel aware mode offers significantly improved error resilience in voice communication over packet-switched networks such as Voice-over-IP (VoIP) and Voice-over-LTE (VoLTE). The error resilience is achieved using a form of in-band forward error correction. Source-controlled coding techniques are used to identify candidate speech frames for bitrate reduction, leaving spare bits for transmission of partial copies of prior frames such that a constant bit rate is maintained. The self-contained partial copies are used to improve the error robustness in case the original primary frame is lost or discarded due to late arrival. Subjective evaluation results from ITU-T P.800 Mean Opinion Score (MOS) tests are provided, showing improved quality under channel impairments as well as negligible impact to clean channel performance.

Super-wideband bandwidth extension for speech in the 3GPP EVS codec

This paper describes the time-domain bandwidth extension (TBE) framework employed to code wideband and super-wideband speech in the newly standardized 3GPP EVS codec. The TBE algorithm uses a nonlinear harmonic modeling technique that incorporates principles of time-domain envelope-modulated noise mixing. At 13.2 kbps, the super-wideband coding of speech uses as low as 1.55 kbps for encoding the spectral content from 6.4-14.4 kHz. Subjective evaluation results from ITU-T P.800 Mean Opinion Score (MOS) tests are provided, showing significantly improved quality compared to the other standardized SWB codecs under both clean speech and speech with background noise.

Two-stage speech/music classifier with decision smoothing and sharpening in the EVS codec

In most internationally recognized standardized multi-mode codecs, signal classification is performed in a single step by either linear discrimination or SNR-based metrics. The speech/music classifier of the EVS codec achieves greater discrimination than these single-step models by combining Gaussian mixture modelling (GMM) with a series of context-based improvement layers. Additionally, unlike traditional GMM classifiers the EVS model adopts a short hangover period, allowing it to track transitions between music and speech. Misclassifications are mitigated by applying a novel decision smoothing and sharpening technique. The results in relatively static environments demonstrate that the new two-stage approach with selective hangover leads to classification accuracies comparable to speech/music classifiers with longer hangovers. They also show that the new approach leads to faster and more accurate switching of coding modes than conventional classifiers for more complex audio environments such as advertisements, jingles and speech superimposed on music.

Low bit rate high-quality MDCT audio coding of the 3GPP EVS standard

This paper presents a low bit-rate MDCT coder, which is adopted as a part of the recently standardized codec for Enhanced Voice Services. To maximize codec performance for NB to SWB input signals for low bit-rates (7.2 to 16.4 kbps), new adaptive bit-allocation and spectrum quantization schemes, which emphasize perceptually important spectrum while efficiently coding full spectrum, was introduced into the low bit-rate MDCT coder. Further, small symbol switched Huffman coding is exploited for reducing the bits consumption for quantizing band energies of the spectrum. Finally, the performance of the coder is illustrated with some listening test results.

Watermark based access control to copyrighted content

One application of digital watermarks deals with control of copying, playback, transmission and other uses of copyrighted content. In this application, the watermark extractor is typically in user-controlled devices, which means that an attacker could mount the oracle attack, or attempt reverse engineering of the watermarking technology. Furthermore, the watermark extractor has to use limited device resources in order to maintain low implementation cost. On the other hand, in this application the payload is relatively small, and watermarks are typically embedded redundantly. In this paper, we describe methods to utilize the embedding redundancy to maximize robustness and security of the system while maintaining low implementation cost. Some of the proposed techniques include iterative extraction process based on initial sparse and random sampling of the distortion space, followed by partial or tentative watermark extraction that triggers extractor adjustment to estimated distortion parameters. Distortion estimation is based on partial watermark matching to pre-distorted templates which can be preselected using experimental approach based on pilot signals. Upon detection of one or more partial or tentative watermarks a new selection of pre-distorted templates is designed to improve distortion resolution and extraction reliability. Extrapolation of watermark segments in the vicinity of partial or tentative watermark extraction is done in view of estimated distortion and an algorithm for combining results of multiple extrapolation segments is proposed. The techniques proposed here allow reduction in processing load and reduction in false positive rates, which in turn can be used to increase watermark robustness to distortion and improved overall security of the system.

Linear prediction based comfort noise generation in the EVS codec

A Discontinuous transmission (DTX) system, which is widely adopted in speech codecs, is an important function for speech communication systems that can reduce the transmission bandwidth by at least a half. Within a DTX system, the comfort noise generation (CNG) plays a key role in the overall quality. Critical performance parameters with respect to the CNG including the transition quality from active to comfort noise (CN) frame, the quality of CN spectrum estimation, wider bandwidth rendering and the DTX efficiency have all been found to be very important. This paper describes a series of new technologies developed for the EVS codec aiming to address the performance of the CNG: A new hangover based CN analysis technique provides improved CNG transition quality. A new entropy based CN spectrum estimation technique and a new hybrid CNG scheme improve the CN spectrum estimation. Finally, a novel bandwidth extension technique for efficient rendering of high-frequency CN and a novel technique improving the DTX efficiency by controlling the DTX hangover length are described.