Codec mode

Description

In 3GPP speech coding, a codec mode refers to a distinct configuration of a speech codec that operates at a specific bitrate and may employ different coding techniques. For the Adaptive Multi-Rate (AMR) codec, there are eight full-rate codec modes, each with a bitrate ranging from 4.75 kbps to 12.2 kbps, plus a low-rate SID (Silence Insertion Descriptor) mode used for comfort noise generation during discontinuous transmission (DTX). Similarly, AMR-Wideband (AMR-WB) has nine codec modes with bitrates from 6.6 kbps to 23.85 kbps. Each mode is optimized for different channel conditions and trade-offs between speech quality and error robustness.

The concept of codec modes is central to the adaptive nature of AMR and AMR-WB. The encoder can switch between modes on a frame-by-frame basis, typically every 20 ms, based on channel quality feedback from the network. This allows the system to maintain acceptable speech quality even as radio conditions degrade, by using lower bitrate modes that are more robust to errors. The network controls which modes are allowed through the Codec Configuration, specifically the Active Codec Set (ACS). The ACS is a subset of all available modes, and the encoder can only switch within this set.

Codec mode adaptation is managed by the network using a combination of in-band signaling and out-of-band control. In UMTS, the Transport Format Combination Indicator (TFCI) carries the codec mode information for each frame. In LTE and NR, the codec mode is signaled via the RTP payload format, which includes a Codec Mode Request (CMR) field that the receiver can use to request a specific mode from the sender. The network can also use explicit signaling, such as the AMR Control (AMR-C) protocol, to command mode changes.

Each codec mode has a unique index (0 to 7 for AMR full-rate modes, 0 to 8 for AMR-WB) that is used in signaling. The mapping of mode indices to bitrates is standardized in 3GPP TS 26.103 for AMR and TS 26.114 for AMR-WB and EVS. For example, AMR mode 7 corresponds to 12.2 kbps, which is the highest quality mode and is also compatible with GSM Enhanced Full Rate (EFR). This interoperability is important for interworking between 3GPP and other systems.

The use of multiple codec modes also enables efficient operation during silence periods. When DTX is enabled, the encoder switches to the SID mode, which transmits comfort noise parameters at a much lower rate (typically 1.8 kbps or less). This reduces interference and saves battery power. The SID mode is not considered a full-rate codec mode but is part of the codec's mode set.

In summary, codec modes are the building blocks of adaptive speech coding in 3GPP systems. They allow the network to balance speech quality and capacity, adapt to changing radio conditions, and ensure interoperability across different technologies. The concept has been extended to newer codecs like EVS, which supports even more modes and higher quality.

Purpose & Motivation

The concept of codec modes was introduced to enable adaptive speech coding in cellular networks, addressing the limitations of fixed-rate codecs. Before AMR, codecs like GSM Full Rate (FR) and Enhanced Full Rate (EFR) operated at a single bitrate, which meant that either speech quality or robustness could be optimized, but not both simultaneously. In varying radio conditions, a fixed-rate codec would either suffer from poor quality (if the bitrate was too low for good quality) or from high error rates (if the bitrate was too high to be protected adequately). The need for a codec that could adapt to channel conditions led to the development of AMR, which introduced multiple codec modes.

The primary motivation for codec modes is to provide a trade-off between speech quality and error resilience. Higher bitrate modes offer better speech quality but are more susceptible to transmission errors, while lower bitrate modes are more robust but offer lower quality. By allowing the network to switch between modes dynamically, AMR can maintain acceptable quality across a wide range of channel conditions. This is particularly important in mobile environments where signal strength and interference fluctuate rapidly.

Historically, the introduction of AMR in 3GPP Release 4/5 (for GSM and UMTS) marked a significant advancement. The concept of codec modes was first defined in the AMR specification, and later extended to AMR-WB in Release 5. The provided data indicates that the term 'Codec mode' was introduced in Release 8, but this likely refers to its inclusion in the 3GPP vocabulary, as the underlying concept existed earlier. The definition in Release 8 was 'for a given channel mode, the bit partitioning between the speech and channel codecs', which reflects the original AMR design where the codec mode determined how bits were split between speech and channel coding. In Release 12, the definition was updated to the current one, emphasizing the bitrate aspect, which aligns with the evolution towards packet-switched voice and the separation of speech and channel coding in LTE.

Codec modes continue to be relevant in modern networks, enabling features like link adaptation and robust voice over IP. They are a key enabler for high-quality voice services in 3GPP systems.