Research in sound discrimination demonstrates that listeners exhibit reduced sensitivity to acoustic differences between allophones, as opposed to phonemes. Previous studies indicates that highly predictable, complementary distribution of allophones contributes to this limited sensitivity by providing strong contextual cues. Building on these insights, this study investigates the role of predictability in sound discrimination within a supervised contrastive learning framework. Specifically, we examine how varying levels of predictability affect the ability to distinguish sounds and whether this influence is categorical or gradual. Additionally, we explore the interaction between acoustic distance and predictability, as well as how the presence of other contrasts within a language modulates this process. Our findings indicate that only full predictability leads to a significant decline in discrimination performance, demonstrating a categorical effect. This impairment can be alleviated as acoustic distance increases. Moreover, the presence of additional contrasts sharing the relevant acoustic dimension enhances discriminability, showing the importance of contextual contrasts in speech perception.

This study uses a modeling approach to explore the development of spectral and positional encodings in speech sounds. Humans rely on their auditory system to differentiate between individual sounds in words by analyzing both spectral properties of phonemes and their relative positions. Previous neuroscientific research has identified specific neural populations in the auditory cortex that respond to spectral processing, while behavioral studies have confirmed humans’ ability to perceive the relative positions of phonemes in speech sequences. To investigate these encodings, a Long Short-Term Memory (LSTM) autoencoder with a cross-attention mechanism trained on Mel-spectrogram transformed from raw speech data is employed in this research. By conducting ABX tests on the model’s representations at various learning stages, we observe the emergence of spectral and positional encodings. The results show that the model excels in distinguishing spectral features similar to neuroscientific findings, and also reveals independent positional encoding through accurate temporal distinctions. Furthermore, we illustrate the developmental trajectory of spectral and positional encodings during the learning process, proposing the need for further investigating their neural correlates.