Segmenting vocalizations into distinct units is essential for interpreting sounds. Songbirds, especially the zebra finch with its hierarchical song structure, provide valuable models for studying neural representations of complex sounds. In this study, we recorded zebra finches' neural activity from the auditory pallium during song processing using a movable electrode array, collecting data from 51 recording sites. We employed stacked bidirectional long short-term memory (BiLSTM) deep neural networks to decode the amplitude envelope and time-locked envelope features of the songs. To evaluate how effectively neural activity segments continuous songs into units and decodes amplitude, the networks were trained using local field potential (LFP) and multi-unit activity envelope (MUAe) data. Our results demonstrate that both amplitude envelope and time-locked features can be accurately decoded from ensemble responses using LFP and MUAe. While the performance of LFP and MUAe was similar, MUAe provided slightly better results for envelope decoding. We observed temporal information is not uniformly present across the auditory pallium. By analyzing the information rate of recorded neurons and the decoding performance of our networks, we identified specific brain regions that code the temporal aspects. Most of these regions are located in the primary auditory area, Field L. Notably, we found a linear correlation between neurons' mean information rate at recording sites and the decoding performance achieved using MUAe. Our high-performance decoding of temporal features illustrates how neural representations facilitate the segmentation of songs. We found that the temporal features of songs are also coded at the single-neuron level and that zebra finches employ region-specific encoding strategies within their auditory processing network. These findings provide valuable insights for future research into the intricate neural processes involved in vocal communication.