PeRDict - Perfect Rhymes Dictionary Database
Source: https://github.com/scrosseye/PeRDict-database
Revision: 25b33897419521a08826d7ccb13f069a816a48d0
File: full_data_elp_perfect_rhymes.csv
SHA-256: f6c9623754d4edc0b677c83b07e1354571d121a9846ae98f9197f4642070e2dd

The shipped JSON preserves all 47,865 CSV records and all five count columns.
Counts: full ELP overlap, and COCA vocabulary bands of 1,000, 2,500, 5,000,
and 10,000 words. The counts are word-level perfect-rhyme neighborhood sizes.
The published CSV does not contain rhyme pairs, pronunciations, or lemma counts.

Afterword uses these counts as a bounded ranking preference for words and phrase
endings with more common perfect-rhyme options. CMUdict continues to supply
pronunciations and determine actual matching words. PeRDict totals are research
dataset counts, not the number of currently displayed or unused suggestions.
PeRDict's definition also excludes identical onsets; differences from Afterword's
current pronunciation inventory and matching rules can produce different totals.
Missing entries are unknown, not zero. A zero is not proof a word cannot rhyme.

Prepared with scripts/import-perdict.py. No upstream code is executed.
No separate license file is present in the source repository at this revision.