Enhanced Voice Services

Description

Enhanced Voice Services (EVS) is a comprehensive audio codec suite standardized by 3GPP, designed as the successor to the AMR-WB codec for IMS-based voice services like VoLTE and VoNR. Its architecture is built around a highly flexible core codec that can operate in multiple modes: a primary Code-Excited Linear Prediction (CELP) mode for traditional speech coding, a Modified Discrete Cosine Transform (MDCT)-based mode for high-quality music and general audio, and a hybrid mode that seamlessly blends both. The codec supports a wide range of sampling rates and bitrates, from 5.9 kbps for super-efficient narrowband operation up to 128 kbps for ultra-high-definition fullband stereo audio. Key components include a robust voice activity detector (VAD), a sophisticated discontinuous transmission (DTX) mechanism for silence compression, and advanced packet loss concealment (PLC) algorithms to mask the effects of frame erasures in packet-switched networks.

In operation, the EVS encoder analyzes the input audio signal, classifies it as speech or music/audio, and selects the optimal coding mode and bitrate based on network conditions and service requirements signaled via the Session Description Protocol (SDP). For speech, the CELP mode models the vocal tract using linear prediction and represents the excitation signal with an adaptive and fixed codebook. For music or mixed content, the MDCT mode provides a transform-domain representation, offering superior audio quality at higher bitrates. The encoded bitstream is packetized into frames, typically 20 ms, and transmitted over the IP network. The decoder reconstructs the audio, applying PLC if packets are lost, using predictive techniques based on previously received frames to generate a smooth, natural-sounding replacement.

EVS plays a critical role in the 3GPP ecosystem by being the mandatory and recommended codec for Voice over LTE (VoLTE) and Voice over NR (VoNR), ensuring interoperability and a consistent high-quality user experience across networks and devices. Its advanced error resilience features, including channel-aware mode switching and frame erasure concealment, make it exceptionally robust against packet loss and jitter, which is vital for maintaining call quality over wireless links. The codec's efficiency allows operators to offer superior voice quality while optimizing spectral efficiency, supporting higher capacity on their networks. Furthermore, its support for super-wideband (50-7000 Hz) and fullband (20-20000 Hz) audio enables immersive voice calls, bridging the quality gap between traditional telephony and streaming media services.

Purpose & Motivation

EVS was created to address the limitations of previous 3GPP voice codecs, primarily AMR-NB and AMR-WB, in the context of all-IP networks like LTE and 5G. While AMR-WB provided a significant quality improvement over narrowband telephony, the evolution towards high-definition multimedia services and the inherent challenges of packet-switched networks (e.g., packet loss, delay jitter) demanded a more robust and higher-fidelity audio solution. The driving motivation was to define a single, future-proof codec that could deliver studio-quality voice, efficiently handle music and other audio content during calls (e.g., on-hold music), and provide unparalleled robustness in poor radio conditions, thereby enhancing the overall user experience for operator-led voice services.

Historically, the transition to VoLTE exposed the need for a codec that could match the quality expectations set by Over-The-Top (OTT) communication apps while meeting the stringent reliability and interoperability requirements of carrier networks. EVS solves the problem of quality degradation under packet loss through its sophisticated error concealment and channel-aware operation, a significant improvement over prior codecs. It also solves the bandwidth inefficiency problem by offering superior audio quality at similar or lower bitrates than AMR-WB, allowing operators to conserve precious radio resources. Its creation was motivated by the desire to make IMS-based voice the unequivocally best voice service available, with quality and features that could not be matched by circuit-switched fallback or OTT alternatives, thereby securing the role of operator services in the IP era.