Speech Expert Group

Description

The Speech Expert Group (SEG) is a specialized technical working group within the 3GPP organizational structure, operating under the Technical Specification Group Services and System Aspects (TSG SA). Its primary mandate is the standardization of speech and audio codecs, as well as the methodologies for testing and assessing their quality and performance in mobile networks. The group's work is critical for ensuring interoperable, high-quality voice and audio services across all 3GPP-defined systems, from GSM and UMTS to LTE and 5G. The SEG does not develop the core codec algorithms itself but rather selects, profiles, and standardizes codecs developed by external bodies (like ITU-T, ETSI, or 3GPP member companies) for use within the 3GPP ecosystem.

The SEG's activities encompass several key areas. First, it specifies the codec bitrates, frame structures, and operational modes that must be supported by terminals and network equipment for mandatory codecs like the Adaptive Multi-Rate (AMR) codec for narrowband speech and the AMR-Wideband (AMR-WB) codec for HD Voice. Second, it defines the detailed protocol for codec negotiation and mode adaptation during call setup and within a call (in-band signaling), which is essential for robust operation over variable radio conditions. Third, the group develops and maintains test specifications for codec performance, including objective and subjective quality measurement methodologies (e.g., using PESQ or POLQA algorithms) and conformance testing procedures to ensure interoperability between different vendors' implementations.

How the SEG works involves a consensus-based standardization process. The group reviews contributions from member companies, evaluates candidate codecs through rigorous testing campaigns (often involving multiple independent labs), and creates the technical specifications that define the codec's 3GPP profile. These specifications cover the codec's interface to the radio access and core network protocols, its adaptation for circuit-switched and packet-switched domains (e.g., in VoLTE and VoNR), and its integration with services like voice over IP and emergency calls. The SEG also addresses emerging audio needs, such as music delivery, multi-channel audio, and immersive audio experiences, expanding its scope beyond traditional telephony speech. Its outputs are integral specifications that ensure a user in a GSM network can have a high-quality voice call with a user on a 5G VoNR connection, using the most appropriate codec negotiated seamlessly between the devices and networks.

Purpose & Motivation

The SEG was formed to address the fundamental requirement for standardized, high-quality digital voice in mobile networks. In the early days of digital cellular (GSM), proprietary codecs would have led to interoperability failures and poor user experience. The group's purpose is to provide a single, agreed-upon set of speech codec standards that all network operators and device manufacturers implement, guaranteeing that voice calls work reliably and sound good across different networks, countries, and device types. It solves the problem of fragmentation and ensures a consistent baseline for voice quality.

Historically, as networks evolved from circuit-switched to packet-switched voice (VoIP), the SEG's role expanded to adapt speech codecs for IP transport, manage packet loss concealment, and define quality of service requirements for voice packets. The motivation for creating and maintaining the SEG includes the continuous drive for improved voice quality (e.g., moving from narrowband to wideband and fullband audio), increased bandwidth efficiency (better quality at lower bitrates), and support for new services like voice over LTE (VoLTE) and voice over NR (VoNR). The group ensures backward compatibility while enabling forward-looking enhancements, balancing quality, complexity, and interoperability in every new release of the 3GPP standards.

Key Features

• Standardization and selection of mandatory and optional speech/audio codecs for 3GPP systems
• Definition of codec negotiation and mode adaptation procedures for circuit and packet-switched domains
• Development of test specifications for codec performance and conformance
• Specification of objective and subjective speech quality assessment methodologies
• Profiling of external codecs (e.g., from ITU-T) for 3GPP network and device implementation
• Evolution of codec standards to support HD Voice, super-wideband, and fullband audio

Evolution Across Releases

Defining Specifications