Composite Source Signal

Description

The Composite Source Signal (CSS) is a precisely defined artificial test signal specified by 3GPP for the objective performance evaluation of speech and audio codecs. It is not a functional component of the live network architecture but a critical tool in the development, verification, and type approval phases. The CSS is designed to simulate a complex, challenging acoustic scenario that stresses codec algorithms, allowing engineers to measure key performance indicators like speech quality, noise robustness, and processing delay under controlled, repeatable laboratory conditions.

The signal itself is a composite waveform generated by mixing multiple audio sources according to a strict recipe defined in the relevant 3GPP test specifications (e.g., TS 26.132, TS 26.907). It typically combines elements such as clean speech samples, background noise of various types (e.g., babble, car, street), and sometimes tonal interferers or music. The specific levels, spectral characteristics, and temporal alignment of these components are standardized to create a 'worst-case' or highly representative test condition. This composite nature is crucial because it moves beyond testing with simple, clean speech signals and evaluates how a codec performs with the complex, noisy audio encountered in real-world usage.

During testing, the CSS is fed as an input to the codec or device under test. The output of the codec's encoding and decoding chain (or the end-to-end transmission path in a system test) is then analyzed. This analysis uses objective perceptual quality measurement algorithms, such as POLQA (Perceptual Objective Listening Quality Analysis) or PESQ (Perceptual Evaluation of Speech Quality), which compare the degraded output signal to the original reference CSS. The resulting scores, like a Mean Opinion Score (MOS), quantify the codec's performance. The role of CSS is to provide the unchanging, high-quality reference against which all these measurements are made, ensuring that test results are comparable across different labs, manufacturers, and test campaigns.

The specifications governing CSS (e.g., TS 26.132 for speech, TS 26.907 for audio) detail its exact composition, digital format, and usage procedures. Compliance with these test specifications, including the correct application of the CSS, is mandatory for network equipment and user devices seeking 3GPP certification. Therefore, while invisible to the end-user, the CSS is a foundational element in the quality assurance ecosystem that guarantees the voice and audio services in 3GPP networks meet minimum performance standards and provide a consistent user experience.

Purpose & Motivation

The Composite Source Signal was created to solve the fundamental problem of inconsistent and non-reproducible codec testing. Before its standardization, different manufacturers and test houses might use their own proprietary test signals or simple tones to evaluate codec performance. This made it impossible to directly compare results between different vendors' equipment or to have confidence that a device certified by one lab would perform adequately in another network. The CSS provides a universal, challenging benchmark that ensures all parties are testing against the same rigorous standard.

Its development was motivated by the need for robust objective testing methodologies as mobile networks evolved and introduced more advanced, and sometimes more complex, speech and audio codecs (like AMR, AMR-WB, and later EVS). These codecs employ sophisticated techniques like noise suppression, discontinuous transmission (DTX), and packet loss concealment. To properly assess these features, a simple clean speech signal is insufficient. The CSS, with its mix of speech and noise, effectively tests the codec's ability to handle real-world acoustic environments, ensuring that quality is maintained not just in quiet offices but also in noisy streets or moving vehicles.

By establishing this common reference signal, 3GPP enabled efficient conformance testing and type approval processes. It allows regulatory bodies and network operators to define clear, measurable pass/fail criteria for audio quality. This, in turn, drives quality improvements across the industry, as codec and device manufacturers must optimize their implementations to perform well when tested with the standardized CSS. It ultimately protects the end-user experience by ensuring a baseline level of audio performance for all certified devices on the network.

Key Features

• Standardized waveform composition for repeatable testing
• Combines multiple audio sources (speech, noise, interferers) to simulate complex acoustic scenes
• Serves as the mandatory reference signal for 3GPP audio codec conformance tests
• Enables objective quality measurement using algorithms like POLQA and PESQ
• Provides a challenging benchmark to stress-test codec robustness and performance
• Ensures comparability of test results across different laboratories and manufacturers

Evolution Across Releases

Defining Specifications