OSCR

Neural rhythms as priors of speech computations.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches
  1. [1] § Methods › Model description ↔ Arabic/coRNN.py, lines 7–46 · score 0.72 · recurrent weight, hidden states, tanh, velocity, bias, activation
  2. [2] § Methods › Network training ↔ Arabic/task.py, lines 66–200 · score 0.58 · cross entropy loss, optimizer, dimension, trained, signal, networks
  3. [3] § Methods › Network training ↔ English/task.py, lines 66–198 · score 0.58 · cross entropy loss, optimizer, dimension, trained, signal, networks

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 75 lines · 2.2 KB · BSD-3-Clause · 1 match

  1. from torch import nn
  2. import torch
  4. device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
  7. class coRNNCell(nn.Module):
  8. def __init__(self, network_type, n_inp, n_hid, dt, gamma, epsilon):
  9. super(coRNNCell, self).__init__()
  11. # network parameters
  12. self.network_type = network_type
  13. self.dt = dt
  14. self.gamma = gamma
  15. self.epsilon = epsilon
  17. # define activation fxn
  18. activation = "tanh"
  19. if activation == "tanh":
  20. self.activation = nn.Tanh()
  21. elif activation == "sin":
  22. self.activation = torch.sin
  23. else:
  24. raise NotImplementedError
  26. # input weights
  27. self.I_ext = nn.Linear(in_features=n_inp, out_features=n_hid, bias=True)
  29. # recurrent weights of hidden states
  30. self.R = nn.Linear(in_features=n_hid, out_features=n_hid, bias=False)
  32. # recurrent weights of velocity of hidden states
  33. self.F = nn.Linear(in_features=n_hid, out_features=n_hid, bias=False)
  35. def forward(self, x, hy, hz):
  37. # our update equations
  38. alpha = 1
  39. activation = self.activation(
  40. alpha * (self.R(hy) + self.F(hz) + self.I_ext(x))
  41. )
  43. hz = hz + self.dt * (activation - self.gamma * hy - self.epsilon * hz)
  44. hy = hy + self.dt * hz
  46. return hy, hz, activation
  49. class coRNN(nn.Module):
  50. def __init__(self, network_type, n_inp, n_hid, n_out, dt, gamma, epsilon):
  51. super(coRNN, self).__init__()
  53. # network parameters
  54. self.n_hid = n_hid
  55. self.gamma = gamma
  56. self.epsilon = epsilon
  57. self.n_out = n_out
  59. self.cell = coRNNCell(network_type, n_inp, n_hid, dt, self.gamma, self.epsilon)
  60. self.readout = nn.Linear(n_hid, n_out)
  62. def forward(self, x):
  64. # initialize hidden states
  65. hy = torch.zeros(x.size(1), self.n_hid, device=device)
  66. hz = torch.zeros(x.size(1), self.n_hid, device=device)
  68. # roll the recurrence forward over the full sequence
  69. for t in range(x.size(0)):
  70. hy, hz, activation = self.cell(x[t], hy, hz)
  72. # read out the final hidden state only
  73. output = self.readout(hy)
  75. return output

coRNN.py at commit 5f3448d, under BSD-3-Clause · at the source

Overview

Authors: Nand Chandravadia1,2, Nabil Imam1,2
ORCID iDs: Nabil Imam
  1. School of Computational Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA
  2. Machine Learning Center, Georgia Institute of Technology, Atlanta, GA 30332, USA
Institutions: Georgia Institute of Technology (United States)
Journal: Patterns (New York, N.Y.), volume 7, issue 8, article 101563
Dates: received 1 November 2025; accepted 21 April 2026; published online 19 May 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1016/j.patter.2026.101563 · PMID 42630811 · PMCID PMC13494615 · OpenAlex W4410205376
Open access: gold, a free copy (OpenAlex)
Preprint: doi:10.1101/2025.05.06.652542
Status: code verified
Categories: cognitive (subfield)
Methods: Connectivity, Preprocessing, Spectral & time-frequency, Machine learning
Keywords: brain rhythms, speech rhythms, inductive biases, coupled oscillators, oscillatory neural networks, wave-RNNs
Topic: Neural Networks and Applications (Artificial Intelligence, Computer Science), according to OpenAlex
Funding: National Science Foundation (2319060, 2223811)
Citations: not cited yet (Europe PMC); 61 references in the paper

Abstract

Endogenous rhythms of auditory neural circuits have a striking resemblance to the temporal modulations of incoming speech signals. Here, we show that these rhythms may serve as priors for speech recognition, encoding knowledge of speech structure in the dynamics of network computations. In a network of coupled oscillators, we find that speech is readily identified when characteristic frequencies of the oscillators match low-frequency circuit rhythms in the auditory cortex. When signal and circuit rhythms are mismatched, speech identification is impaired. Compared to a baseline recurrent neural network without intrinsic oscillations, the coupled oscillatory network has significantly higher performance in speech recognition across languages but not in the recognition of signals that lack speech-like structure, such as urban sounds. Our results suggest a central computational role of brain rhythms in speech processing.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repositories

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

nandchandravadia/rhythmRNNs_speech

License: BSD-3-Clause
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Commit: 5f3448d85312bdbdc541a3816d19485110b22472, 22 May 2026
Languages: Python (16)
Size: 23 files, 16 scripts
Software Heritage: not archived
Found in: “Data and code availability”
Holds: README, license file
Not found: CITATION.cff, environment file, tests, continuous integration, documentation
Tools: PyTorch (16 files), pandas (4 files)
Availability: 1 check, the latest on 28 September 2026: the link answers
  • 28 September 2026: the link answers
18 files
At the source: github.com/nandchandravadia/rhythmRNNs_speech

Zenodo 19411357

License: CC-BY-4.0
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Size: 2 files, 0 scripts
Software Heritage: not checked
Found in: “Data and code availability”
Holds: README, license file
Not found: CITATION.cff, environment file, tests, continuous integration, documentation
Availability: 1 check, the latest on 28 September 2026: the link answers (HTTP 200)
  • 28 September 2026: the link answers (HTTP 200)
2 files
At the source: doi.org/10.5281/zenodo.19411357

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 2 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 16 scripts, each with its path and the digest of its content;
  • 3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data and code availability

The code to reproduce the results of this study is available as a GitHub repository: https://github.com/nandchandravadia/rhythmRNNs_speech. The code is also available at a Zenodo repository: https://doi.org/10.5281/zenodo.19411357.61

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 2, 28 September 2026

  • Authors: added Nabil Imam (0000-0003-2143-2286); removed Nabil Imam
  • Funding: added National Science Foundation: 2319060, 2223811

Version 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 2 authors, 6 keywords, 59 references.

Cite

This paper

Chandravadia, N., & Imam, N. (2026). Neural rhythms as priors of speech computations. Patterns (New York, N.Y.), 7(8), 101563. https://doi.org/10.1016/j.patter.2026.101563

BibTeX

@article{chandravadia2026neural,
author = {Chandravadia, Nand and Imam, Nabil},
title = {{Neural rhythms as priors of speech computations}},
journal = {Patterns (New York, N.Y.)},
year = {2026},
month = may,
volume = {7},
number = {8},
pages = {101563},
publisher = {Elsevier},
issn = {2666-3899},
doi = {10.1016/j.patter.2026.101563},
url = {https://doi.org/10.1016/j.patter.2026.101563},
pmid = {42630811},
pmcid = {PMC13494615}
}

RIS

TY - JOUR
AU - Chandravadia, Nand
AU - Imam, Nabil
TI - Neural rhythms as priors of speech computations
T2 - Patterns (New York, N.Y.)
J2 - Patterns (N Y)
PY - 2026
DA - 2026/05/19
VL - 7
IS - 8
SP - 101563
SN - 2666-3899
PB - Elsevier
DO - 10.1016/j.patter.2026.101563
UR - https://doi.org/10.1016/j.patter.2026.101563
LA - en
ER -

CSL-JSON

{
"id": "10.1016/j.patter.2026.101563",
"type": "article-journal",
"title": "Neural rhythms as priors of speech computations",
"container-title": "Patterns (New York, N.Y.)",
"author": [
{
"family": "Chandravadia",
"given": "Nand"
},
{
"family": "Imam",
"given": "Nabil"
}
],
"container-title-short": "Patterns (N Y)",
"volume": "7",
"issue": "8",
"page": "101563",
"DOI": "10.1016/j.patter.2026.101563",
"PMID": "42630811",
"PMCID": "PMC13494615",
"ISSN": "2666-3899",
"publisher": "Elsevier",
"URL": "https://doi.org/10.1016/j.patter.2026.101563",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
19
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1126/sciadv.aec6112 [code]
Dual computational systems in the development and evolution of mammalian brains.
Journal: Science advances
In common: 3 references, author Nabil Imam
[2] doi:10.1371/journal.pbio.3003924
Endogenous auditory and motor brain rhythms predict individual speech tracking.
Journal: PLoS biology
In common: cognitive, 5 references
[3] doi:10.1162/imag.a.1305
Can you <i>feel</i> what I am saying? Speech-based vibrotactile stimulation enhances the cortical tracking of attended speech in a multi-talker background.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: cognitive, 4 references
[4] doi:10.1523/eneuro.0379-25.2026 [code]
Next-Generation Neural Mass Models Reproduce Features of Speech Processing.
Journal: eNeuro
In common: 4 references
[5] doi:10.1038/s41598-026-57256-0 [code]
Pupil- and gaze dynamics track emotion content in naturalistic speech.
Journal: Scientific reports
In common: cognitive, 3 references
[6] doi:10.1002/aur.70312 [code]
Aberrant Neural Entrainment to Word-Level Speech Patterns in Fragile X Syndrome: Evidence for a Statistical Learning Deficit.
Journal: Autism research : official journal of the International Society for Autism Research
In common: cognitive, 3 references
[7] doi:10.1162/imag.a.1348 [code]
The role of high-amplitude bursts of high-gamma activity in naturalistic speech and music listening.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: pandas, cognitive, 2 references
[8] doi:10.7554/elife.105482 [code]
Principles of gamma synchrony predict figure-ground perception in texture stimuli.
Journal: eLife
In common: pandas, 2 references
[9] doi:10.7554/elife.100056 [code]
Multi-talker speech comprehension at different temporal scales in listeners with normal and impaired hearing.
Journal: eLife
In common: PyTorch, pandas, cognitive, 1 reference
[10] doi:10.1038/s41467-026-72931-6 [code]
Three parsimonious spatiotemporal patterns in cerebellum reveal individual traits in function and behavior.
Journal: Nature communications
In common: PyTorch, pandas, cognitive, 1 reference

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.