Both humans and songbirds acquire speech and song respectively, in a critical period early in development. Interestingly, remarkable parallels can be found between the establishment of these complex behaviors making songbirds, zebra finches in particular, a valid and translational model to study aspects of human speech learning [1]. Zebra finches learn their song during two well-defined partly-overlapping sensitive periods early in development i.e. the sensory (20-40 days post hatching (dph); memorization of tutor song) and sensorimotor phase (35-90 dph; trial-and-error practicing to match own vocalizations to the tutor song). Until recently, zebra finch brain development could only be studied by invasive methods such as histology and molecular testing, which strictly impedes with the establishment of causal relationships between a specific improvement in behavioral performance and the underlying biological substrate. Here we use in vivo Diffusion Tensor Imaging (DTI) to establish a spatiotemporal profile of structural brain changes which might help unveil brain areas implicated in the process of song learning.Juvenile male (Taeniopygia guttata; n=16) and female (n=18) zebra finches were scanned on a 7 T Bruker MR system, equipped with a 400 mT/m gradient insert, at seven time points crucial for the process of vocal learning i.e. at 20 (initiation sensory phase), 30 (mid sensory phase), 40 (end sensory beginning sensorimotor phase), 65 (mid sensorimotor phase), 90 (end sensorimotor phase), 120 (crystallization, adulthood) and 200 (‘advanced’ adulthood) dph. The birds were anaesthetized with isoflurane (induction 2.5%; maintenance: 1.2-1.4%). Each imaging session consisted of a DTI scan (TR: 7000 ms; TE: 22 ms; δ: 4 ms; Δ: 12 ms; b: 670 s/mm²; 60 diffusion gradient directions; spatial resolution: (0.19x0.19x0.24)cm³; 28 horizontal slices; 2 repetitions) followed by a T2-weighted 3D anatomical RARE (TR: 2500 ms; TE: 11 ms; RARE factor: 8; zerofilled to a matrix of (228x256x142) with voxel resolution (0.07x0.07x0.07)cm³). Throughout the entire imaging procedure, respiration rate and temperature were kept within narrow physiological ranges (40.0 ± 0.2)°C. All datasets were spatially normalized to a population-based template after which a whole-brain voxel-wise repeated measures ANOVA was performed on the smoothed Fractional Anisotropy, Mean Diffusion, Axial Diffusion and Radial Diffusion maps. The songs of juvenile male birds were recorded the first two days after each imaging session starting from the 65 dph time point so as to determine the exact phase of vocal learning and quantify the similarity to the tutor song.Figure 1 shows a statistical map illustrating the clusters which display a significant change in fractional anisotropy over the different time points, in male zebra finches. Clusters can be co-localized with several components of the song control circuitry e.g. HVC (1), arcopallium (including the robust nucleus of the arcopallium (RA); 4) and tracts e.g. tractus occipitomesencephalicus (OM; 6), lamina mesopallialis (LaM; 3), lamina frontalis superior (LFS; 2), Area X surroundings (5) etc. Closer examination of the difference between sensory (20-30 dph) and sensorimotor phase (40-65 dph) in male birds for FA, highlights several brain areas in control of the motoric aspect of singing i.e. HVC, arcopallium (containing RA), the tractus OM which is the highway tract connecting the caudal motor pathway (HVC-RA) to the syrinx (avian homologue of the larynx). This can be related to the animals’ behavior: male birds only start to vocalize during the sensorimotor phase. Consequently, the data suggest that while juvenile birds are practicing and trying to match their vocalizations to the tutors’ song, the brain circuitry underlying vocal motor production is being formed. Statistical testing of the mean diffusion maps, shows two main clusters over time i.e. the biggest changes are found between 20-30 dph and 40-65 dph (Fig. 2). Interestingly, no significant changes in mean diffusion could be observed between 30-40 dph and from 65 dph onwards. Fascinatingly, when comparing 20 versus 30 dph the peak difference of the cluster resembles the shape of Area X, a brain region known to be involved in vocal exploration necessary for trial-and-error learning.