pith:GMH4UAMF
PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization
PairAlign generates compact audio token sequences by training each view's output to be likely under the other's encoder while contrasting unrelated examples.
arxiv:2605.06582 v2 · 2026-05-07 · cs.LG · cs.CL · cs.SD
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{GMH4UAMFDXWDWTBEAYXTDM7KOR}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
On TIMIT retrieval, it preserves edit-distance search while reducing archive token count by 55%. A continuous-sweep probe shows lower local overlap than a dense geometric tokenizer, but stronger length control and bounded edit trajectories under 100 ms shifts.
That optimizing cross-view sequence likelihood with unrelated negatives as contrast produces token sequences whose edit-distance properties generalize to downstream tasks without direct supervision on those properties.
PairAlign learns compact audio token sequences via self-alignment of paired content views using an autoregressive decoder, achieving strong cross-view consistency and edit-distance preservation while reducing token count by 55% on TIMIT.
Receipt and verification
| First computed | 2026-06-09T02:07:28.588832Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
330fca01851dec3b4c24062f31b3ea745f2547c7a9d3c564477ee7730ed16628
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/GMH4UAMFDXWDWTBEAYXTDM7KOR \
| jq -c '.canonical_record' \
| python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 330fca01851dec3b4c24062f31b3ea745f2547c7a9d3c564477ee7730ed16628
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "7b681725430bf02f2e528276b1172415e2a4c1fc77a25c71711f1db6a0af57e6",
"cross_cats_sorted": [
"cs.CL",
"cs.SD"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.LG",
"submitted_at": "2026-05-07T17:11:22Z",
"title_canon_sha256": "59807b35e32d4a73ebd341180dead6b1f03cafad47f37df62a6354572ca93743"
},
"schema_version": "1.0",
"source": {
"id": "2605.06582",
"kind": "arxiv",
"version": 2
}
}