Ultra-Low Bitrate speech Codec

Description

The Ultra-Low Bitrate speech Codec (ULBC) is a speech coding technology standardized in 3GPP Release 20, specified in technical report 26.940. It is designed to operate at bitrates significantly lower than existing codecs like AMR-WB or EVS, targeting scenarios where bandwidth is extremely scarce or costly. ULBC achieves this through advanced compression techniques, likely leveraging machine learning or parametric coding models to synthesize intelligible speech from minimal data. The codec is intended for applications such as satellite communication, deep coverage IoT voice, or emergency services in remote areas.

Architecturally, ULBC fits into the 3GPP speech codec framework as an alternative codec within the IMS or MCPTT (Mission Critical Push-to-Talk) subsystems. It can be negotiated via SDP (Session Description Protocol) using standard codec negotiation mechanisms. The codec operates on 20 ms frames, typical for 3GPP speech codecs, but with a much smaller payload size. This reduces the required radio resources and backhaul bandwidth, enabling voice over networks that would otherwise be data-only. The codec may also include forward error correction (FEC) and packet loss concealment (PLC) mechanisms to maintain quality under high latency and loss conditions.

Key components of ULBC include the encoder, which analyzes speech signals and extracts essential parameters, and the decoder, which reconstructs the speech using those parameters. The encoder likely employs a source-filter model or a neural network-based vocoder to achieve ultra-low bitrates. The decoder may use a similar model to generate the output waveform. The codec also defines a bit-exact reference implementation to ensure interoperability across vendors. Performance requirements, such as mean opinion score (MOS) and latency, are specified in 26.940 to guide implementation.

In the network, ULBC is relevant for services that require voice but operate under extreme constraints. For example, in NB-IoT or eMTC, where data rates are limited, ULBC could enable voice over the cellular IoT. It could also be used in 5G satellite access (NTN) to provide voice services in remote areas. The codec's ultra-low bitrate also makes it suitable for storing voice messages or for machine-to-machine voice communication. As 3GPP moves towards 6G, ULBC may become a foundational codec for ubiquitous voice coverage.

ULBC is specified in 3GPP TR 26.940, which provides the codec's technical description, test vectors, and performance characterization. The codec is expected to be integrated into future releases of specifications like TS 26.114 (IMS Multimedia Telephony) and TS 26.179 (Mission Critical Push-to-Talk). Its development reflects the ongoing need for efficient voice coding in diverse network environments, from terrestrial IoT to non-terrestrial networks.

Purpose & Motivation

ULBC was created to address the growing need for voice communication in networks where bandwidth is extremely limited or expensive. Traditional codecs like AMR-NB (4.75–12.2 kbps) and EVS (5.9–128 kbps) are not efficient enough for scenarios such as satellite IoT, where every bit counts. The motivation was to enable voice services in these constrained environments without sacrificing intelligibility.

Previous approaches to low-bitrate coding, such as MELP (2.4 kbps) or Codec2 (700 bps), were not standardized within 3GPP and lacked the robustness and interoperability required for cellular networks. ULBC fills this gap by providing a 3GPP-standardized solution that can be integrated into existing IMS and MCPTT architectures. It also supports the vision of ubiquitous voice coverage, extending cellular voice to remote areas and new device types.

The development of ULBC was driven by the expansion of 3GPP technologies into non-traditional areas, including satellite communication (NTN) and massive IoT. In these domains, voice is often an afterthought, but there are use cases like emergency calls, remote maintenance, and mission-critical communications that require it. ULBC provides a standardized, efficient way to deliver voice in these contexts, ensuring interoperability and quality.