Multilevel modeling of between-speaker and within-speaker variation in spontaneous speech tempo | Synapse