Music and speech rhythms are hierarchically organized, including grouping beats to create metrical structures. Previously, we showed that infants can be primed via loudness accents to interpret a metrically ambiguous (unaccented) rhythm either in duple meter (groupings of 2 beats) or in triple meter (groupings of 3 beats), as measured by larger mismatch responses (MMRs) in electroencephalographic recordings for the perceptually strong compared with weak beat in the unaccented rhythm [Flaten, E., Marshall, S. A., Dittrich, A., & Trainor, L. J. Evidence for top-down meter perception in infancy as shown by primed neural responses to an ambiguous rhythm. European Journal of Neuroscience, 55, 2003-2023, 2022]. Given that infants primed with a duple or triple metrical interpretation heard the same ambiguous stimulus at test, this indicated top-down meter perception. The effects were stronger in the duple-primed infants, although this may have reflected that the stimulus was also slightly biased toward the duple meter. Here, we investigated the generalizability of 6-month-old infants' top-down meter processing by varying the tempo of the rhythm from priming to test. We also used an isochronous test rhythm to ensure there was no duple or triple bias in the stimulus. Results showed that infants' MMRs were not enhanced for deviants on primed strong versus weak beat positions; however, infants taking regular music classes who were primed with triple meter showed a larger MMR for beat 5 (strong beat for duple) than beat 4. Furthermore, duple-primed infants tracked the rhythm more strongly than triple-primed infants, as shown by steady-state evoked potentials. These results suggest that, although infants did not show evidence of generalizing metrical priming across varying tempi, a bias for duple metrical interpretation develops early and may be accelerated by participation in music classes.