SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.01855 · ICML · 2022

Self-Supervised Learning with Random-Projection Quantizer for Speech Recognition

Jiahui Yu, Chung-Cheng Chiu, Yonghui Wu, Yu Zhang, James Qin

arXiv · PDF · Open in the Atlas

Code that ran

12 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 4 violated it and 4 have no contract verdict; 3 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2202.01855.json · badge: /api/ran/2202.01855.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
lucasnewman/best-rq-pytorch community (classifier label)
classifier basis not recorded on the link
10 ran · 1 no recorded run
wenet-e2e/wenet community
Papers with Code archive
2 ran · 0 no recorded run
HarunoriKawano/BEST-RQ community
Papers with Code archive
0 ran · 2 no recorded run
FunctionStatusWhere it lives
Wav2vecGumbelVectorQuantizer Ran
ran on a synthesized input
wenet-e2e/wenet/wenet/models/ssl/wav2vec2/quantizer.py
code served (permissive licence) · get_code("b1ed164ff39acfaf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

INIT_ARGS = {
    "features_dim": 4,
    "num_codebooks": 2,
    "num_embeddings": 8,
    "embedding_dim": 8,
    "hard": False,
}

ARGS = {
    "input": torch.randn(2, 3, 4),
    "input_mask": torch.ones(2, 3, dtype=torch.bool),
    "temperature": 1.0,
}

if __name__ == "__main__":
    from typing import Tuple
    
    def gumbel(shape: torch.Size, dtype: torch.dtype, device: torch.device):
        return -torch.log(-torch.log(torch.empty(shape, device=device).uniform_(torch.finfo(dtype).tiny, 1.0)))

    class Wav2vecGumbelVectorQuantizer(torch.nn.Module):

        def __init__(self, features_dim: int=256, num_codebooks: int=2, num_embeddings: int=8192, embedding_dim: int=16, hard: bool=False) -> None:
            super().__init__()
            self.num_groups = num_codebooks
            self.num_codevectors_per_group = num_embeddings
            assert embedding_dim % num_codebooks == 0.0
            self.embeddings = torch.nn.parameter.Parameter(torch.empty(1, num_codebooks * num_embeddings, embedding_dim // num_codebooks), requires_grad=True)
            torch.nn.init.uniform_(self.embeddings)
            self.weight_proj = torch.nn.Linear(features_dim, num_codebooks * num_embeddings)
            self.hard = hard

        @staticmethod
        def _compute_perplexity(probs, mask=None):
            if mask is not None:
                mask_extended = torch.broadcast_to(mask.flatten()[:, None, None], probs.shape)
                probs = torch.where(mask_extended.to(torch.bool), probs, torch.zeros_like(probs))
                marginal_probs = probs.sum(dim=0) / mask.sum()
            else:
                marginal_probs = probs.mean(dim=0)
            perplexity = torch.exp(-torch.sum(marginal_probs * torch.log(marginal_probs + 1e-07), dim=-1)).sum()
            return perplexity

        def forward(self, input: torch.Tensor, input_mask: torch.Tensor, temperature: float=1.0) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
            b, t, _ = input.size()
            hidden = self.weight_proj(input)
            hidden = hidden.reshape(b * t * self.num_groups, -1)
            if not self.hard:
                gumbels = gumbel(hidden.size(), hidden.dtype, hidden.device)
                codevector_probs = torch.nn.functional.softmax((hidden + gumbels) / temperature, dim=-1)
                codevector_soft_dist = torch.nn.functional.softmax(hidden.reshape(b * t, self.num_groups, -1), dim=-1)
                perplexity = self._compute_perplexity(codevector_soft_dist, input_mask)
            else:
                codevector_idx = hidden.argmax(axis=-1)
                codevector_probs = torch.nn.functional.one_hot(codevector_idx, hidden.shape[-1]) * 1.0
                codevector_probs = codevector_probs.reshape(b * t, self.num_groups, -1)
                perplexity = self._compute_perplexity(codevector_probs, input_mask)
            targets_idx = codevector_probs.argmax(-1).reshape(b, t, -1)
            codevector_probs = codevector_probs.reshape(b * t, -1)
            codevectors_per_group = codevector_probs.unsqueeze(-1) * self.embeddings
            codevectors = codevectors_per_group.reshape(b * t, self.num_groups, self.num_codevectors_per_group, -1)
            codevectors = codevectors.sum(-2).reshape(b, t, -1)
            return (codevectors, perplexity, targets_idx)

    model = Wav2vecGumbelVectorQuantizer(**INIT_ARGS)
    output = model(**ARGS)
    print("Output shapes:", output[0].shape, output[1].shape, output[2].shape)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
calc_same_padding Ran
ran, violated the contract we drafted
lucasnewman/best-rq-pytorch/best_rq_pytorch/conformer.py
code served (permissive licence) · get_code("a41cb31b42e919b1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

ARGS = {
    'kernel_size': 5
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cast_tuple Ran
ran, honoured the contract we drafted
lucasnewman/best-rq-pytorch/best_rq_pytorch/trainer.py
code served (permissive licence) · get_code("5cfcc1d3230706ff")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function cast_tuple takes a single parameter 't' and returns it as a tuple if it's not already a tuple or list
# We need to test different input types to demonstrate the function works

# Test case 1: A scalar value
test_scalar = 42

# Test case 2: A tensor
test_tensor = torch.randn(2, 4)

# Test case 3: A list
test_list = [1, 2, 3, 4]

# Test case 4: A tuple
test_tuple = (5, 6, 7)

# Test case 5: A numpy array
test_numpy = np.array([1, 2, 3])

# We'll use a scalar as the main test case
ARGS = {
    "t": 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
curtail_to_shortest_collate Ran
ran on a synthesized input
lucasnewman/best-rq-pytorch/best_rq_pytorch/data.py
code served (permissive licence) · get_code("88639cb78b5cf76a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 4, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Test case 1: Single batch of tensors (not tuples)
# The decorator will detect is_one_data=True and return (result,)
single_tensor_batch = [
    torch.randn(5, 4),
    torch.randn(3, 4),
    torch.randn(7, 4),
]

# Test case 2: Multiple batches (tuples of tensors)
# The decorator will detect is_one_data=False and process each position
multi_tensor_batch = [
    (torch.randn(5, 4), torch.randn(5, 3)),
    (torch.randn(3, 4), torch.randn(3, 3)),
    (torch.randn(7, 4), torch.randn(7, 3)),
]

ARGS = {
    "data": single_tensor_batch
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
beartype, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 4, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
default Ran
ran, violated the contract we drafted
lucasnewman/best-rq-pytorch/best_rq_pytorch/best_rq.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("60fff7c3c400d7ff")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ThomasMrY/VCT, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

exists Ran
ran, violated the contract we drafted
lucasnewman/best-rq-pytorch/best_rq_pytorch/attend.py
pointer only (licence: MIT); the stored copy get_code serves, from ThomasMrY/VCT, is pointer only (licence: NONE) · get_code("aa5486a3650902d8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ThomasMrY/VCT, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_mask_subset_prob Ran
ran on a synthesized input
lucasnewman/best-rq-pytorch/best_rq_pytorch/best_rq.py
code served (permissive licence) · get_code("c2d0191b3cd5f3db")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a mask tensor with shape (batch=2, seq=4)
# mask should be boolean, True where we want to consider positions
mask = torch.tensor([
    [True, True, True, False],
    [True, True, False, False]
], dtype=torch.bool)

# prob should be a float between 0 and 1
prob = 0.5

# min_mask should be a non-negative integer
min_mask = 0

ARGS = {
    "mask": mask,
    "prob": prob,
    "min_mask": min_mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_optimizer Ran
ran on a synthesized input
lucasnewman/best-rq-pytorch/best_rq_pytorch/optimizer.py
code served (permissive licence) · get_code("9354ff16b3840873")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned AdamW. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from torch.optim import AdamW, Adam

torch.manual_seed(0)

# Create a simple model to extract parameters from
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        # 2D parameters (should be weight decayable)
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
        # 1D parameters (should not be weight decayable)
        self.bias_param = nn.Parameter(torch.randn(2))
        # 0D parameter (scalar, should not be weight decayable)
        self.scalar_param = nn.Parameter(torch.tensor(0.5))

model = SimpleModel()

# Get parameters from model
params = list(model.parameters())

# Verify we have parameters with different dimensions
# linear1.weight: [3, 4] (2D)
# linear1.bias: [3] (1D)
# linear2.weight: [2, 3] (2D)
# linear2.bias: [2] (1D)
# bias_param: [2] (1D)
# scalar_param: [] (0D)

ARGS = {
    'params': params,
    'lr': 0.0001,
    'wd': 0.01,
    'betas': (0.9, 0.99),
    'eps': 1e-08,
    'filter_by_requires_grad': False,
    'group_wd_params': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type AdamW — a AdamW output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a AdamW output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
gumbel Ran
ran; our contract draft was wrong, not the code
wenet-e2e/wenet/wenet/models/ssl/wav2vec2/quantizer.py
code served (permissive licence) · get_code("aa69d004af2aa82f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    "shape": torch.Size([2, 4]),
    "dtype": torch.float32,
    "device": torch.device("cpu"),
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
once Ran
ran; our contract draft was wrong, not the code
lucasnewman/best-rq-pytorch/best_rq_pytorch/attend.py
code served (permissive licence) · get_code("52e10a7d6b0fa036")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned function. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from functools import wraps

def once(fn):
    called = False

    @wraps(fn)
    def inner(x):
        nonlocal called
        if called:
            return
        called = True
        return fn(x)
    return inner


def sample_function(x):
    """A simple function that the decorator will wrap."""
    return x * 2


ARGS = {
    "fn": sample_function
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
separate_weight_decayable_params Ran
ran; our contract draft was wrong, not the code
lucasnewman/best-rq-pytorch/best_rq_pytorch/optimizer.py
pointer only (licence: MIT); the stored copy get_code serves, from levyisthebest/echopulse_prelease, is pointer only (licence: NONE) · get_code("fdc11aefc97f3abb")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from levyisthebest/echopulse_prelease, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

yes_or_no Ran
ran, violated the contract we drafted
lucasnewman/best-rq-pytorch/best_rq_pytorch/trainer.py
code served (permissive licence) · get_code("d95562c398a6b25e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import unittest.mock as mock

# The function uses input() to get user input and checks if the answer is in ('yes', 'y')
# We need to mock input() to return a valid answer

ARGS = {
    'question': 'Do you want to continue?'
}

# Mock the input function to return 'y' so the function can execute
import builtins
original_input = builtins.input
builtins.input = mock.Mock(return_value='y')
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Config Not yet run
unverified (harvested, no recorded run)
HarunoriKawano/BEST-RQ/model/random_projection_quanzier.py
code served (permissive licence) · get_code("88a745959f70f9ed")
How it ran

no invocation recorded.

RandomProjectionQuantizer Not yet run
unverified (harvested, no recorded run)
HarunoriKawano/BEST-RQ/model/random_projection_quanzier.py
code served (permissive licence) · get_code("c55d1b0032cf0be6")
How it ran

no invocation recorded.

collate_one_or_multiple_tensors Not yet run
unverified (harvested, no recorded run)
lucasnewman/best-rq-pytorch/best_rq_pytorch/data.py
code served (permissive licence) · get_code("ef413882b65418df")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 9 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 2 the same month, 9 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
exists 486 other papers
78 dated earlier, 4 the same month, 401 later, 1 same year, order not determinable, 2 undated · 78 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1611.07004 (2016-11)
Image-to-Image Translation with Conditional Adversarial Networks
1804.04637 (2018-04)
EMBER: An Open Dataset for Training Static PE Malware Machine Learning Models
and 483 more not named here
default 163 other papers
38 dated earlier, 2 the same month, 122 later, 1 undated · 27 only as an assembly brick
1603.08155 (2016-03)
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
1810.00825 (2018-10)
Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks
1907.01470 (2019-07)
Augmenting Self-attention with Persistent Memory
and 160 more not named here
once 19 other papers
1 dated earlier, 18 later · 4 only as an assembly brick
2006.11239 (2020-06)
Denoising Diffusion Probabilistic Models
2203.07852 (2022-03)
Block-Recurrent Transformers
2205.14135 (2022-05)
FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
and 16 more not named here
separate_weight_decayable_params 8 other papers
8 later
2301.11325 (2023-01)
MusicLM: Generating Music From Text
2302.04761 (2023-02)
Toolformer: Language Models Can Teach Themselves to Use Tools
2303.05511 (2023-03)
Scaling up GANs for Text-to-Image Synthesis
and 5 more not named here
yes_or_no 7 other papers
7 later
2301.11325 (2023-01)
MusicLM: Generating Music From Text
2302.01327 (2023-02)
Dual PatchNorm
2305.09636 (2023-05)
SoundStorm: Efficient Parallel Audio Generation
and 4 more not named here
calc_same_padding 4 other papers
1 dated earlier, 3 later
2005.08100 (2020-05)
Conformer: Convolution-augmented Transformer for Speech Recognition
2203.15149 (2022-03)
CMGAN: Conformer-based Metric GAN for Speech Enhancement
2310.01889 (2023-10)
Ring Attention with Blockwise Transformers for Near-Infinite Context
and 1 more not named here
cast_tuple 4 other papers
4 later
2302.01327 (2023-02)
Dual PatchNorm
2305.09636 (2023-05)
SoundStorm: Efficient Parallel Audio Generation
2306.15687 (2023-06)
Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
and 1 more not named here
get_optimizer 2 other papers
2 later
2306.15687 (2023-06)
Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
2501.04926 (2025-01)
FLowHigh: Towards Efficient and High-Quality Audio Super-Resolution with Single-Step Flow Matching
collate_one_or_multiple_tensors 1 other paper
1 later
2306.15687 (2023-06)
Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present a simple and effective self-supervised learning approach for speech recognition. The approach learns a model to predict the masked speech signals, in the form of discrete labels generated with a random-projection quantizer. In particular the quantizer projects speech inputs with a randomly initialized matrix, and does a nearest-neighbor lookup in a randomly-initialized codebook. Neither the matrix nor the codebook is updated during self-supervised learning. Since the random-projection quantizer is not trained and is separated from the speech recognition model, the design makes the approach flexible and is compatible with universal speech recognition architecture. On LibriSpeech our approach achieves similar word-error-rates as previous work using selfsupervised learning with non-streaming models, and provides lower word-error-rates and latency than wav2vec 2.0 and w2v-BERT with streaming models. On multilingual tasks the approach also provides significant improvement over wav2vec 2.0 and w2v-BERT.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2202.01855")
get_code_for_paper("2202.01855")
have("2202.01855")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2202.01855.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2202.01855.svg)](https://syntology.ai/paper/2202.01855)

Connect an agent — have() is free.