Pith Number

pith:QSKAJ67D

pith:2025:QSKAJ67DCH2K5MRRRTKUWQNYGF

not attested not anchored not stored refs resolved

AaSP: Aliasing-aware Self-Supervised Pre-Training for Audio Spectrogram Transformers

Kohei Yamamoto, Kosuke Okusa

AaSP uses input-estimated kernels to fuse high-frequency cues lost to aliasing in audio spectrogram pre-training.

arxiv:2512.03637 v2 · 2025-12-03 · cs.SD · cs.LG · stat.ML

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{QSKAJ67DCH2K5MRRRTKUWQNYGF}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Under fine-tuning, the full AaSP framework achieves state-of-the-art results on AS-20K, ESC-50, and NSynth among compared self-supervised baselines, while remaining competitive elsewhere.

C2weakest assumption

That the aliasing introduced by convolutional patchification with temporal downsampling is a primary performance bottleneck and that fusing features from alias-prone modulation bands via an input-estimated kernel will integrate useful high-frequency cues without introducing new instabilities or overfitting to the estimation process.

C3one line summary

AaSP learns aliasing-stable audio representations by augmenting patch tokens with adaptive subband features from alias-prone bands and using teacher-student masked modeling plus multi-mask contrastive regularization, reaching SOTA on AS-20K, ESC-50, and NSynth under fine-tuning.

References

45 extracted · 45 resolved · 3 Pith anchors

[1] Bert: Pre- training of deep bidirectional transformers for language understanding, 2019

[2] Dinov2: Learning robust visual features without supervision, 2023

[3] Masked autoencoders that listen, 2022

[4] MAE-AST: Masked Autoencoding Audio Spectrogram Transformer, 2022

[5] SSAST: Self-Supervised Audio Spectrogram Transformer, 2022

Receipt and verification

First computed	2026-05-17T23:39:16.928935Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

849404fbe311f4aeb2318cd54b41b83177f1058cb606d3c3f6065a2e5380f23b

Aliases

arxiv: 2512.03637 · arxiv_version: 2512.03637v2 · doi: 10.48550/arxiv.2512.03637 · pith_short_12: QSKAJ67DCH2K · pith_short_16: QSKAJ67DCH2K5MRR · pith_short_8: QSKAJ67D

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF \
  | 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: 849404fbe311f4aeb2318cd54b41b83177f1058cb606d3c3f6065a2e5380f23b

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "35cb764cb3457dc05a500bd8f4fa4042cc9f06e2cc683b35baa12ca56633f960",
    "cross_cats_sorted": [
      "cs.LG",
      "stat.ML"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.SD",
    "submitted_at": "2025-12-03T10:17:35Z",
    "title_canon_sha256": "459598cee3f40ec87ef99209c8389e007e070721f92715f7544fc1d810b56584"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2512.03637",
    "kind": "arxiv",
    "version": 2
  }
}