This Recommendation specifies a subjective test methodology for evaluating the quality of speech communication systems that include noise suppression algorithms, and more generally for speech-in-noise conditions. The methodology employs separate rating scales to independently assess three perceptual dimensions: the speech signal (SIG), the background noise (BAK), and the overall quality (OVRL). By requiring listeners to attend successively to each dimension, the method reduces response uncertainty and improves rating reliability compared with single-scale approaches such as Absolute Category Rating (ACR). It is applicable to both terminal-based and offline-simulated test configurations.