Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Distilling Spectrograms into Tokens: Fast and Lightweight Bioacoustic Classification for BirdCLEF+ 2025

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A lightweight pipeline that turns Mel-spectrograms into discrete tokens and learns static skip-gram embeddings from unlabeled soundscapes can classify 206 species in roughly 6 minutes of CPU time, beating chance but remaining far below…

desk verdict Honest BirdCLEF working note: useful CPU benchmarks and a clean tokenization baseline, but the 'viability' claim overstates a 0.52 private AUC without a token-free ablation on the competition metric. read the letter →

arxiv 2507.08236 v1 pith:7QCYTK5X submitted 2025-07-11 cs.SD eess.AS

Anthony Miyaguchi , Murilo Gustineli , Adrian Cheung This is my paper
classification cs.SDeess.AS
keywords bioacousticsspectrogramtokenizationself-supervisedlearningefficientinferenceacousticmonitoringBirdCLEFskip-gramembeddingsstatic
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that bioacoustic soundscapes can be classified under a strict 90-minute CPU-only deadline in two complementary ways: by compiling pre-trained models to run on CPUs, and by a novel pipeline called the Spectrogram Token Skip-Gram (STSG) that turns Mel-spectrograms into discrete tokens and learns static embeddings from unlabeled audio using a skip-gram objective. The authors' key result for STSG is that it achieves a public ROC-AUC of 0.559 and a private 0.520 on the competition's 206-species soundscape classification task while running in a projected 6 minutes, about four times faster than their best transfer-learning baseline, which reached 0.810 public but took about 26 minutes. If this is right, it demonstrates that fast tokenization with static embeddings is a viable route for resource-constrained acoustic monitoring, even though the accuracy gap to pre-trained models remains large.

What carries the argument

The central object is the STSG pipeline. Spectrograms are computed with 768 Mel bands at 8 frames per second, normalized, projected with PCA (128 components retain 87% of variance), and quantized by K-means into a 16,384-token codebook. A skip-gram negative-sampling model (embedding dimension 1024, context window 80, uniform negative sampling) is trained on the token sequences and produces a static lookup table; classification averages token embeddings per 5-second window and passes them through a linear layer. The skip-gram objective is what carries the argument: tokens that co-occur near each other in time receive nearby vectors, so the averaged embeddings encode the acoustic content of a window without any labeled data. The paper also relies on model compilation to make pre-trained backbones CPU-viable, which sets the baseline that STSG is compared against.

What would settle it

Run the STSG v2.1 configuration with 768 Mel bands, 128 PCA components, 16,384 K-means clusters, embedding dimension 1024, context window 80, and negative-sampling exponent 0.0 on the official test set, measuring macro ROC-AUC on 5-second intervals and total CPU time: if the score is not meaningfully above 0.500 or the runtime exceeds the 90-minute deadline on the specified CPU, the paper's viability claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that treating bioacoustic audio as a sequence of discrete spectrogram tokens, rather than as a spectrogram image, yields a classifier accurate enough to beat chance and fast enough for extreme edge constraints. The STSG pipeline clusters PCA-reduced Mel-spectrogram frames into a 16,384-token vocabulary, learns static skip-gram embeddings from token co-occurrence in unlabeled soundscapes, averages the embeddings over 5-second windows, and runs a linear classification head. On the competition's leaderboard it reports ROC-AUC 0.559 public / 0.520 private, with a projected ~6 minutes of CPU time for 700 minutes of soundscapes, about four times faster than the best-scoring pre-trained baseline (0.810 public) and about three times faster than the compiled pre-trained bird-song model (0.729 public). The authors take this as evidence that fast tokenization with static embeddings is a viable alternative in resource-constrained settings, while acknowledging that STSG trails transfer learning by a wide accuracy margin and that a student-teacher distillation from the stronger model actually hurt performance.

Load-bearing premise

The surrogate validation task—a single-label, whole-recording classification across 20 hand-picked species—faithfully represents the real multi-label 5-second soundscape task across 206 species, even though scores on the surrogate saturate and the real leaderboard scores are far lower.

Editorial extensions

If this is right

  • Pre-trained bioacoustic backbones can be compiled for CPU with roughly a 10x speedup, making transfer learning practical under a 90-minute CPU-only deadline.
  • The best compiled backbone scores ROC-AUC 0.810 public / 0.778 private, while the compiled bird-song model scores 0.729 / 0.711.
  • STSG classifies the same test set in a projected 6 minutes at ROC-AUC 0.559 public / 0.520 private, outperforming a plain Mel-spectrogram average (surrogate F1 0.12 vs 0.56) and beating chance.
  • Hyperparameter choices matter: a 16,384-token vocabulary and PCA-reduced (128-dim) Mel-spectrogram input improved surrogate F1, and lowering the negative-sampling exponent helped when subsampling was less aggressive.
  • Distilling teacher logits into the STSG student model reduced surrogate F1 from 0.56 to 0.47, so the current embedding space does not yet absorb the teacher's knowledge.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The 6-minute CPU runtime is projected rather than measured on the competition instance; an editor's inference is that the 90-minute deadline headroom would allow a more powerful tokenizer or a small transformer to replace the K-means tokens without leaving the intended use case.
  • The gap between surrogate F1 (around 0.56) and leaderboard ROC-AUC (0.52-0.56) suggests the surrogate task is much easier; a natural extension is to evaluate STSG directly on multi-label 5-second frames, using class-balanced sampling or per-species token adjustments to lift ROC-AUC.
  • The authors' finding that static embeddings from temporal co-occurrence help classification suggests similar token-plus-embedding pipelines could apply to other under-resourced audio monitoring domains, such as insect or amphibian surveys, where labeled data are scarce.
  • Since the student-teacher distillation failed, the more promising direction implied but not tested is to use the token vocabulary as a discrete input to a lightweight transformer, or to derive static embeddings from a large pretrained model via the compression approach the authors cite for future work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This working note describes the DS@GT BirdCLEF+ 2025 submission, which combines transfer learning from pretrained bioacoustic backbones with a newly proposed lightweight pipeline called Spectrogram Token Skip-Gram (STSG). STSG converts audio into Mel-spectrograms, reduces frames with PCA, quantizes them via Faiss K-means, learns static token embeddings with Word2Vec skip-gram, averages the embeddings over 5-second windows, and trains a linear classifier. The paper reports strong transfer-learning baselines (BirdSetEfficientNetB1: 0.810 public / 0.778 private ROC-AUC; Perch with TFLite: 0.729 / 0.711), and a fast STSG variant with a projected six-minute CPU inference time that reaches 0.559 public / 0.520 private ROC-AUC. The central claim is that this result demonstrates the viability of fast tokenization approaches with static embeddings for bioacoustic classification. Validation is performed on a surrogate single-label whole-recording task over 20 species, and the final scores are from the Kaggle leaderboard.

Significance. If the central claim is supported, the paper would provide a useful data point for CPU-constrained bioacoustic classification: a clustering-plus-static-embedding pipeline that runs in minutes and exceeds chance on a difficult multi-taxon task. The paper has clear strengths: the transfer-learning results are credible and useful baselines; the TFLite speedup for Perch is quantified; the STSG pipeline is described in enough detail to reproduce; hyperparameter sweeps and timing measurements are reported; and the authors release code. I also see no circularity in the evaluation: the STSG embeddings are learned unsupervised, the classification head is trained on surrogate labels, and the leaderboard scores are external. However, the central viability claim is currently confounded with a non-token feature representation (PCA-reduced Mel-spectrogram averaging), and the surrogate validation task is a weak proxy for the competition metric. These issues are load-bearing and require revision.

major comments (3)
  1. [§3.2, §4.3, Tables 9-11] The central claim that STSG demonstrates the viability of tokenization is confounded by the absence of a token-free ablation on the competition metric. The STSG classifier is a linear head over per-window averaged embeddings, where the input features are PCA-reduced Mel-spectrogram frames (Sections 3.2.1 and 3.2.2). The only non-token comparison, the MelSpec Baseline in Table 9, is evaluated on the surrogate single-label F1 task (0.12 vs. 0.56), not on the competition's macro-averaged ROC-AUC. Because the STSG private score is 0.520, only modestly above chance, the observed result could be carried by the averaged PCA features plus the linear head even if the tokenizer and Word2Vec stage contributed nothing. The authors should add a token-free ablation that keeps the same PCA features, the same averaging, the same linear head, and the same test protocol, and reports the competition metric. Without such an ablation, the paper can claim a fast spectrogram-averaging baseline but not a viable tokenization approach.
  2. [§3.3 and Tables 9-10] The surrogate validation task is a weak proxy for the actual competition, and all STSG design choices are made on it. The surrogate is a single-label, whole-recording classification task over 20 species, while the competition is multi-label classification of 5-second intervals with macro-averaged ROC-AUC. The paper itself notes that surrogate ROC-AUC saturates near 1.0 for transfer-learning models and therefore uses F1, which is not the competition metric. Hyperparameters such as vocabulary size, PCA dimensions, and Word2Vec settings are tuned on this proxy, so the justification of these choices depends on the proxy's validity. The authors should either validate the proxy on the competition metric (e.g., by showing that surrogate rankings predict leaderboard rankings for several configurations) or explicitly reframe the paper's contribution as a fast CPU baseline rather than a fully validated bioacoustic tokenizer.
  3. [Table 9 vs. Table 10] There is an unexplained internal inconsistency in the surrogate results: Table 9 reports STSG (v1) macro F1 as 0.45, while Table 10 reports 0.381 for the same model. The micro F1 also differs (0.53 vs. 0.394). If these numbers come from different validation splits or training runs, this must be stated; otherwise the surrogate evaluation is not auditable. This matters because the surrogate is used to justify hyperparameter decisions and to compare STSG variants, so a consistent reporting protocol is needed.
minor comments (5)
  1. [Table 8 and Appendix Table 12] The vector_size sweep lists a value of 1028, which appears to be a typo for 1024; the same typo appears in the appendix.
  2. [Table 11 caption] The caption describes the runtimes as projections for a hypothetical 700-soundscape test set, while the text presents the public and private scores as final leaderboard results. Please clarify which numbers are measured and which are extrapolated, and how the projection was computed.
  3. [§3.2.3] The temperature scaling description says the teacher probabilities are softened by a squared factor of temperature T=3; the exact formula (e.g., logits divided by T^2 versus T) should be stated explicitly.
  4. [§4.1, Figure 5b] The discussion of overfitting in the Word2Vec validation curves is qualitative; reporting a single early-stopping criterion or a selected epoch number would make the procedure reproducible.
  5. [Throughout] The name 'RanaSierraCNN' in Table 1 differs from 'RanaSierraeCNN' in later tables; please standardize.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: STSG embeddings are learned unsupervised, the classification head is trained on surrogate labels, and the leaderboard scores are external evidence. The only self-citation is non-load-bearing.

full rationale

The STSG pipeline is self-contained rather than circular. Mel-spectrogram frames are tokenized by Faiss K-means on PCA-reduced features (Sections 3.2.1–3.2.2), and token embeddings are learned unsupervised with Word2Vec skip-gram; no target label or competition score enters the tokenizer or embedding objective. The linear classification head is trained on the surrogate single-label task (Section 3.3), while the final ROC-AUC scores in Table 11 come from the external Kaggle leaderboard. Hyperparameters are swept on the surrogate F1-macro and then frozen before leaderboard evaluation, so the leaderboard result is not a fitted value. The 'MelSpec Baseline' in Table 9 provides a non-token comparison on the surrogate task, and although no token-free ablation on the competition metric is reported, the absence of an ablation is a confounding/evidence weakness, not a circular reduction by construction. The only self-citation, reference [6] from the same team's 2024 working note, is used to report that neural codecs were previously found non-viable; this is not load-bearing for the STSG derivation. No equation in the paper defines a claimed prediction in terms of the fitted parameter, and no uniqueness theorem or ansatz is imported from the authors' prior work. The finding is therefore no significant circularity, with the modest score reflecting only the non-load-bearing self-citation.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The central claims rest on several domain assumptions (surrogate validity, transferability of pretrained embeddings, information content of K-means tokens) plus a large set of hyperparameters tuned on a small validation set. No new physical entities are introduced.

free parameters (7)
  • K-means vocabulary size = 16,384
    Chosen by sweeping F1 on the 20-species surrogate task; larger vocab (32k) improves F1 but increases training time.
  • PCA dimensions for spectrogram frames = 128
    Keeps 87% variance of 768 Mel bands; selected because lower dimensionality improved validation F1 and memory use.
  • Word2Vec vector_size = 1024 (v2.0-v2.2)
    Swept over 128-1028; larger vectors improved F1 by up to 0.018, with 1.7x training time increase.
  • Word2Vec window = 80
    Context window for skip-gram; default 80 was near-optimal in sweeps, smaller window hurt F1 by 0.013.
  • ns_exponent = 0.0 (v2.1)
    Negative sampling exponent; best value depended on subsampling rate, 0.0 chosen for final STSG model after sweeps.
  • subsampling sample = 1e-5
    Token subsampling threshold; 1e-5 chosen for v2.0 and v2.1, while v2.2 used 1e-4 with ns_exponent -0.75.
  • student-teacher temperature = 3
    Temperature scaling for teacher logits in distillation; chosen by hand, no sweep reported.
assumptions (4)
  • domain assumption The surrogate task based on single-label classification of the training species dataset reflects the multi-label soundscape competition task.
    Invoked in Section 3.3; all hyperparameter tuning and model comparisons rest on this proxy, but the competition uses 5-second multi-label ROC-AUC while the surrogate uses whole-recording F1.
  • domain assumption Pre-trained bioacoustic backbone embeddings transfer to the domain-shifted Middle Magdalena soundscapes.
    Stated in Section 3.1 and supported by cited literature on BirdNET and Perch transfer; underlies all transfer-learning baselines.
  • domain assumption K-means clusters of PCA-reduced Mel-spectrogram frames form semantically meaningful discrete audio tokens.
    Core to STSG; the paper shows clusters yield above-chance classification but does not establish token interpretability.
  • standard math Skip-gram co-occurrence statistics over time encode species-relevant information.
    Standard Word2Vec assumption; the paper's results suggest it holds weakly for this audio domain.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Distilling Spectrograms into Tokens: Fast and Lightweight Bioacoustic Classification for BirdCLEF+ 2025." pith.science (2026). https://pith.science/paper/7QCYTK5X

@misc{pith2026250708236,
  author       = {Pith},
  title        = {Pith review of: Distilling Spectrograms into Tokens: Fast and Lightweight Bioacoustic Classification for BirdCLEF+ 2025},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7QCYTK5X}},
  note         = {Machine review of arXiv:2507.08236}
}
read the original abstract

The BirdCLEF+ 2025 challenge requires classifying 206 species, including birds, mammals, insects, and amphibians, from soundscape recordings under a strict 90-minute CPU-only inference deadline, making many state-of-the-art deep learning approaches impractical. To address this constraint, the DS@GT BirdCLEF team explored two strategies. First, we establish competitive baselines by optimizing pre-trained models from the Bioacoustics Model Zoo for CPU inference. Using TFLite, we achieved a nearly 10x inference speedup for the Perch model, enabling it to run in approximately 16 minutes and achieve a final ROC-AUC score of 0.729 on the public leaderboard post-competition and 0.711 on the private leaderboard. The best model from the zoo was BirdSetEfficientNetB1, with a public score of 0.810 and a private score of 0.778. Second, we introduce a novel, lightweight pipeline named Spectrogram Token Skip-Gram (STSG) that treats bioacoustics as a sequence modeling task. This method converts audio into discrete "spectrogram tokens" by clustering Mel-spectrograms using Faiss K-means and then learns high-quality contextual embeddings for these tokens in an unsupervised manner with a Word2Vec skip-gram model. For classification, embeddings within a 5-second window are averaged and passed to a linear model. With a projected inference time of 6 minutes for a 700-minute test set, the STSG approach achieved a final ROC-AUC public score of 0.559 and a private score of 0.520, demonstrating the viability of fast tokenization approaches with static embeddings for bioacoustic classification. Supporting code for this paper can be found at https://github.com/dsgt-arc/birdclef-2025.

Figures

Figures reproduced from arXiv: 2507.08236 by the authors.

Figure 1
Figure 1. Overview of various experiments as a flow diagram. We utilize pre-trained bioacoustic classifiers and transfer learning onto the BirdCLEF+ task using a surrogate classification task. We also experiment with tokenized MFCC embeddings and student-teacher pre-training on top of these embeddings [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. PaCMAP projection showing the distribution of Perch embeddings for the 10 most frequent species: compau, grekis, roahaw, whtdov, laufal1, trokin, yercac1, trsowl, banana, and wbwwre1. We selected species based on observation count and visualized to assess clustering behavior in the embedding space. graph needed for training can be pruned, fused, and optimized for inference. We achieve an order of magnitude speedup f… view at source ↗
Figure 3
Figure 3. The Spectrogram Token Skip-Gram (STSG) Pipeline. Mel-spectrograms are generated from the source audio, which are then clustered using K-Means to create discrete tokens. A Word2Vec skip-gram model then learns a continuous vector embedding for each token in an unsupervised manner. In an optional pre-training stage, a student model uses these embeddings to produce output logits. A larger teacher model, Perch, also prod… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: F1 score and training time vs. tokenizer vocabulary size. A vocabulary of 16k clusters provides the best performance before diminishing returns [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Hyperparameter tuning and final model performance. (a) F1 score as a function of the negative sampling exponent (ns exponent) for two different subsampling rates. The optimal exponent shifts from 0.0 to -0.75 when using less aggressive subsampling (1e-4). (b) Validatio…
Figure 6
Figure 6. Figure 6: PaCMAP visualizations of bioacoustic model zoo embedding spaces using the training dataset where every point represents an audio track. The color represents the one-dimensional embedding of the audio track in BirdNET space, computed by averaging all embeddings generate…
Figure 7
Figure 7. Figure 7: F1 score and training time vs. tokenizer vocabulary size. A vocabulary of 16k clusters provides the best performance before diminishing returns [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Hyperparameter tuning and final model performance. (a) Tuning the negative sampling exponent shows a clear peak at 1.5. (b) The final tuned model consistently outperforms the baseline [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Mixture-Constrained Max Pooling Improves Separation-Based Bird Species Classification

    eess.AS 2026-07 conditional novelty 5.5 of 10

    An ensemble of MixIT-trained separators with mixture-constrained max pooling improves multi-species bird classification by capturing true-positive gains while suppressing false-positive gains from separation artifacts.

  2. Can Tokens Compete? Token Representations against Supervised CNN Backbones for BirdCLEF+ 2026

    cs.SD 2026-07 conditional novelty 5.0 of 10

    For BirdCLEF+ 2026, a frozen Perch-v2 probe plus a trained HGNetV2-B0 SED net and non-bird prototype heads reach private LB 0.936, while WavTokenizer codec tokens collapse and four general audio transformers lag under...

Reference graph

Works this paper leans on

20 extracted references · 12 canonical work pages · cited by 2 Pith papers

  1. [1]

    J. S. Cañas, S. Kahl, T. Denton, M. P. Toro-Gómez, S. Rodriguez-Buritica, J. L. Benavides-Lopez, J. S. Ulloa, P. Caycedo-Rosales, H. Klinck, H. Glotin, H. Goëau, W.-P. Vellinga, R. Planqué, A. Joly, Overview of BirdCLEF+ 2025: Multi-taxonomic sound identification in the middle magdalena valley, colombia, in: Working Notes of CLEF 2025 - Conference and Lab...

    work page 2025

  2. [2]

    Picek, S

    L. Picek, S. Kahl, H. Goëau, L. Adam, et al., Overview of lifeclef 2025: Challenges on species presence prediction and identification, and individual animal identification, in: International Conference of the Cross-Language Evaluation Forum for European Languages, Springer, 2025

    work page 2025

  3. [3]

    Ghani, T

    B. Ghani, T. Denton, S. Kahl, H. Klinck, Global birdsong embeddings enable superior transfer learning for bioacoustic classification, Scientific Reports 13 (2023) 22876

    work page 2023

  4. [4]

    Williams, B

    B. Williams, B. van Merriënboer, V. Dumoulin, J. Hamer, A. B. Fleishman, M. McKown, J. Munger, A. N. Rice, A. Lillis, C. White, et al., Using tropical reef, bird and unrelated sounds for superior transfer learning in marine bioacoustics, Philosophical Transactions B 380 (2025) 20240280

    work page 2025

  5. [5]

    S. Kahl, T. Denton, H. Klinck, V. Ramesh, V. Joshi, M. Srivathsa, A. Anand, C. Arvind, H. CP, S. Sawant, V. V. Robin, H. Glotin, H. Goëau, W.-P. Vellinga, R. Planqué, A. Joly, Overview of BirdCLEF 2024: Acoustic identification of under-studied bird species in the western ghats, Working Notes of CLEF 2024 - Conference and Labs of the Evaluation Forum (2024)

    work page 2024

  6. [6]

    Miyaguchi, A

    A. Miyaguchi, A. Cheung, M. Gustineli, A. Kim, Transfer learning with pseudo multi-label birdcall classification for ds@gt birdclef 2024, Working Notes of CLEF 2024 - Conference and Labs of the Evaluation Forum (2024)

    work page 2024

  7. [7]

    Zeghidour, A

    N. Zeghidour, A. Luebs, A. Omran, J. Skoglund, M. Tagliasacchi, Soundstream: An end-to- end neural audio codec, CoRR abs/2107.03312 (2021). URL: https://arxiv.org/abs/2107.03312. arXiv:2107.03312

    arXiv 2021

  8. [8]

    Défossez, J

    A. Défossez, J. Copet, G. Synnaeve, Y. Adi, High fidelity neural audio compression, arXiv preprint arXiv:2210.13438 (2022)

    arXiv 2022

Show all 20 references
  1. [9]

    Baevski, Y

    A. Baevski, Y. Zhou, A. Mohamed, M. Auli, wav2vec 2.0: A framework for self-supervised learning of speech representations, Advances in neural information processing systems 33 (2020) 12449– 12460

    2020

  2. [10]

    Mikolov, K

    T. Mikolov, K. Chen, G. Corrado, J. Dean, Efficient estimation of word representations in vector space, arXiv preprint arXiv:1301.3781 (2013)

    2013 arXiv

  3. [11]

    Borsos, R

    Z. Borsos, R. Marinier, D. Vincent, E. Kharitonov, O. Pietquin, M. Sharifi, D. Roblek, O. Teboul, D. Grangier, M. Tagliasacchi, N. Zeghidour, Audiolm: A language modeling approach to audio generation, IEEE/ACM Trans. Audio, Speech and Lang. Proc. 31 (2023) 2523–2533. URL: http...

    2023

  4. [12]

    S. Lapp, T. Rhinehart, L. Freeland-Haynes, J. Khilnani, A. Syunkova, J. Kitzes, Opensoundscape: an open-source bioacoustics analysis package for python, Methods in Ecology and Evolution 14 (2023) 2321–2328

    2023

  5. [13]

    Davis, P

    S. Davis, P. Mermelstein, Comparison of parametric representations for monosyllabic word recognition in continuously spoken sentences, IEEE Transactions on Acoustics, Speech, and Signal Processing 28 (1980) 357–366. doi:10.1109/TASSP.1980.1163420

    1980

  6. [14]

    Douze, A

    M. Douze, A. Guzhva, C. Deng, J. Johnson, G. Szilvasy, P.-E. Mazaré, M. Lomeli, L. Hosseini, H. Jégou, The faiss library (2024). arXiv:2401.08281

    2024 arXiv

  7. [15]

    Řehůřek, P

    R. Řehůřek, P. Sojka, Software Framework for Topic Modelling with Large Corpora, in: Proceedings of the LREC 2010 Workshop on New Challenges for NLP Frameworks, ELRA, Valletta, Malta, 2010, pp. 45–50. http://is.muni.cz/publication/884893/en

    2010

  8. [16]

    Caselles-Dupré, F

    H. Caselles-Dupré, F. Lesaint, J. Royo-Letelier, Word2vec applied to recommendation: Hyperpa- rameters matter, in: Proceedings of the 12th ACM Conference on Recommender Systems, 2018, pp. 352–356

    2018

  9. [17]

    Rauch, R

    L. Rauch, R. Schwinger, M. Wirth, R. Heinrich, D. Huseljic, M. Herde, J. Lange, S. Kahl, B. Sick, S. Tomforde, et al., Birdset: A large-scale dataset for audio classification in avian bioacoustics, arXiv preprint arXiv:2403.10380 (2024)

    2024 arXiv

  10. [18]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, I. Polosukhin, Attention is all you need, Advances in neural information processing systems 30 (2017)

    2017

  11. [19]

    Tulkens, T

    S. Tulkens, T. van Dongen, Model2vec: Fast state-of-the-art static embeddings (2024). URL: https://github.com/MinishLab/model2vec

    2024

  12. [20]

    URL: http://www

    PACE, Partnership for an Advanced Computing Environment (PACE), 2017. URL: http://www. pace.gatech.edu. A. STSG v1 Hyperparameter Tuning The first version of STSG uses a spectrogram with 128 Mel bands and 20 MFCCs. Figure 7: F1 score and training time vs. tokenizer vocabulary ...

    2017

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.