Update README.md

README.md

@@ -175,7 +175,13 @@ distribution.
 
 Overlap is not concentrated at one difficulty. Mixtures span the full range from
 clean (no overlap) to fully overlapped, with the harder end growing as speaker count
-increases (2-4 spk), which is the expected and desired behaviour.
+increases (2-4 spk), which is the expected and desired behavior.
+
+**Definition.** Utterances are placed sequentially, and a single overlap ratio
+`r ∈ {0, 0.25, 0.5, 0.75, 1.0}` is drawn per mixture. For each adjacent speaker pair
+`(A, B)`, the overlap duration is `r × min(dur_A, dur_B)` — i.e. `r` is the fraction of
+the **shorter** of the two adjacent utterances that overlaps. The chosen `r` is stored
+as `overlapped_ratio` in each mixture's JSON.
 
 <img src="./figure/overlap_distribution.png" width="70%">