Comfort Noise Addition

Description

Comfort Noise Addition (CNA) is a sophisticated signal processing function defined within the 3GPP framework, specifically in the context of media handling for voice and audio services. Its primary role is to manage the audio experience during periods of discontinuous transmission (DTX), where the active speech encoder stops transmitting packets during silence to conserve network bandwidth and device battery. Without CNA, these silent gaps would be perceived as complete, unnatural silence or 'dead air,' which can be jarring to users and may even lead to false call disconnection assumptions.

Architecturally, CNA is implemented within the media processing path, typically at the receiving endpoint or within a media gateway or application server that handles voice traffic. It operates in conjunction with voice activity detection (VAD) and silence suppression mechanisms. When VAD indicates the onset of a silent period and DTX is active, the regular speech frames cease. The CNA function then takes over, generating and injecting a comfort noise signal into the audio output stream. This noise is not random white noise but is carefully parameterized to match the characteristics of the background acoustic environment present during the last active speech segment, ensuring a seamless and natural auditory transition from speech to silence and back.

The technical implementation involves analyzing the spectral properties of the background noise during active talk spurts. Parameters such as noise level and spectral shape are extracted and can be transmitted to the receiver via low-bandwidth Silence Insertion Descriptor (SID) frames or similar signaling, as defined in codecs like Adaptive Multi-Rate (AMR) and Enhanced Voice Services (EVS). Alternatively, in more advanced or network-based implementations, the CNA algorithm at the receiver can independently estimate or generate appropriate comfort noise based on previously received audio. The generated noise is typically a shaped, low-level noise that mimics the ambient sound, preventing the abrupt switch to absolute silence.

CNA's role in the network is crucial for Quality of Experience (QoE) in voice over LTE (VoLTE), Voice over NR (VoNR), and other packet-switched voice services. It is a key component in the end-to-end audio chain, working alongside core network elements like the IP Multimedia Subsystem (IMS) and policy controls that manage media optimization. By maintaining a consistent acoustic backdrop, CNA reduces listener fatigue, prevents confusion about call status, and supports the natural flow of conversation, which is essential for user acceptance of efficient but disruptive technologies like DTX.

Purpose & Motivation

CNA was created to solve a fundamental psychoacoustic problem in digital voice communications: the unnatural perception of absolute silence. Early digital voice systems, especially those employing efficient codecs with DTX to save bandwidth, would completely mute audio output during pauses. This created an eerie, 'on-off' soundscape that was fatiguing and could make users think the call had dropped. The purpose of CNA is to fill these artificial silences with a low-level, comfort-matching background noise, thereby preserving the call's natural feel and continuity.

The historical context stems from the evolution of voice codecs and network efficiency measures. As cellular networks advanced from 2G to 4G and 5G, the use of packet-switched voice (VoIP) and sophisticated codecs like AMR-WB and EVS became standard. These codecs heavily utilize DTX for power and bandwidth savings. However, the limitations of previous approaches were clear: either DTX was not used, wasting resources, or it was used without effective comfort noise, degrading perceived quality. CNA provides the necessary compromise, enabling the economic benefits of DTX without sacrificing user comfort.

Furthermore, CNA addresses challenges in varied acoustic environments. Without it, switching between active speech and total silence can make background noise seem to 'pop' in and out, which is particularly distracting in environments with constant low-level noise (e.g., offices, cars). By generating a consistent noise floor, CNA masks these transitions. Its standardization in 3GPP Release 18, as part of broader media enhancement specifications, was motivated by the need for a unified, high-quality voice experience across diverse devices and network conditions, especially critical for operator-grade VoLTE/VoNR services competing with over-the-top applications.

Key Features

• Generates artificial background noise during DTX silent periods
• Uses parameters derived from the active speech's ambient noise spectrum
• Ensures a seamless and natural auditory transition from speech to silence
• Prevents user perception of 'dead air' or call drop
• Works in conjunction with Voice Activity Detection (VAD) and Silence Insertion Descriptor (SID) frames
• Enhances Quality of Experience (QoE) for packet-switched voice services like VoLTE and VoNR

Evolution Across Releases

Defining Specifications