SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1703.09452 · 2017

SEGAN: Speech Enhancement Generative Adversarial Network

arXiv · PDF · Open in the Atlas

Code that ran

26 of 39 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; 3 honoured the contract we drafted, 1 violated it and 4 have no contract verdict; 18 failures of our instrument; 13 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/1703.09452.json · badge: /api/ran/1703.09452.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
santi-pdp/segan canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
santi-pdp/segan_pytorch community (classifier label)
classifier basis not recorded on the link
13 ran · 1 no recorded run
haozh7109/SEGAN-TensorFlow2 community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
andreser09/segan community (classifier label)
classifier basis not recorded on the link
2 ran · 1 no recorded run
asadullah797/segan_impr community (classifier label)
classifier basis not recorded on the link
2 ran · 1 no recorded run
YosukeSugiura/SEGAN reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
develooper1994/MasterThesis reimplementation (classifier label)
classifier basis: classifier_2025
0 ran · 3 no recorded run
huylenguyen806/vnasr community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
ssarfjoo/improvedsegan community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
sufengniu/segan_multi community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
FunctionStatusWhere it lives
KLD Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/ops.py
code served (permissive licence) · get_code("c8cdd110f18d46c9")
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 small tensors with distinct shapes for each parameter
# mean_p: mean of distribution p
mean_p = torch.randn(2, 4)

# std_p: standard deviation of distribution p
std_p = torch.abs(torch.randn(2, 4)) + 0.1  # ensure positive

# mean_g: mean of distribution g
mean_g = torch.randn(2, 4)

# std_g: standard deviation of distribution g
std_g = torch.abs(torch.randn(2, 4)) + 0.1  # ensure positive

ARGS = {
    "mean_p": mean_p,
    "std_p": std_p,
    "mean_g": mean_g,
    "std_g": std_g,
}
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
build_norm_layer Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/modules.py
code served (permissive licence) · get_code("b9130696e49c3339")
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 BatchNorm1d. 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.nn.utils.spectral_norm import spectral_norm

torch.manual_seed(0)

# Create a simple linear layer to use as param for spectral_norm
param_layer = nn.Linear(4, 3)

ARGS = {
    'norm_type': 'bnorm',
    'param': param_layer,
    'num_feats': 5
}
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 BatchNorm1d — a BatchNorm1d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BatchNorm1d 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
collate_fn Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/datasets/se_dataset.py
code served (permissive licence) · get_code("82ea88d96fdaa51d")
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 list. 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 torch.utils.data.dataloader import default_collate

torch.manual_seed(0)

# The collate_fn expects a batch parameter which is a list of samples
# Each sample is a tuple where:
# - sample[0] is the utterance name (a string)
# - sample[1:] are the data elements to be collated

# Create a small batch with 2 samples
# Each sample has: (uttname, tensor1, tensor2)
sample1 = ("utt_001", torch.randn(4, 3), torch.randn(2, 5))
sample2 = ("utt_002", torch.randn(4, 3), torch.randn(2, 5))

batch = [sample1, sample2]

ARGS = {
    "batch": batch
}
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 list — a list 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: none by the harness; any seed is in the setup block
de_emph Ran
ran on a synthesized input
santi-pdp/segan/data_loader.py
code served (permissive licence) · get_code("f97b465794902557")
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 ndarray, ndim 1, dtype family float, shape (5,) 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 numpy as np

# de_emph expects:
# - y: a 1D numpy array (the function accesses y.shape[0] and y[n])
# - coeff: a numeric value (default 0.95)

# Create a small 1D numpy array for y
y = np.array([0.1, 0.2, 0.3, 0.4, 0.5], dtype=np.float32)

ARGS = {
    "y": y,
    "coeff": 0.95
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (5,) 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
denormalize_wave_minmax Ran
ran, honoured the contract we drafted
santi-pdp/segan_pytorch/segan/utils.py
code served (permissive licence) · get_code("d81dd81a9f083f86")
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)

# The function denormalize_wave_minmax takes a single parameter 'x'
# which should be a tensor (based on the mathematical operations performed on it)
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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
emphasis Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
haozh7109/SEGAN-TensorFlow2/Utility/Py_data_Preprocess_Hao.py
code served (permissive licence) · get_code("1056ff58d881d749")
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 ndarray, ndim 3, dtype family float, shape (2, 3, 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 numpy as np

# Create a small batch of signals with shape [batch, channel, time]
# batch=2, channel=3, time=4
signal_batch = np.array([
    [
        [0.1, 0.2, 0.3, 0.4],
        [0.5, 0.6, 0.7, 0.8],
        [0.9, 1.0, 1.1, 1.2]
    ],
    [
        [0.2, 0.3, 0.4, 0.5],
        [0.6, 0.7, 0.8, 0.9],
        [1.0, 1.1, 1.2, 1.3]
    ]
], dtype=np.float32)

ARGS = {
    "signal_batch": signal_batch,
    "emph_coeff": 0.95,
    "pre": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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
emphasis_librosa Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
haozh7109/SEGAN-TensorFlow2/Utility/Py_data_Preprocess_Hao.py
code served (permissive licence) · get_code("cae4d9938745ed11")
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 ndarray, ndim 3, dtype family float, shape (2, 3, 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 numpy as np
import scipy.signal

# Create signal_batch with shape [batch_size, num_channels, time_steps]
# Using batch_size=2, num_channels=3, time_steps=4
signal_batch = np.random.randn(2, 3, 4).astype(np.float32)

ARGS = {
    "signal_batch": signal_batch,
    "emph_coeff": 0.95,
    "pre": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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
flip Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/modules.py
pointer only (licence: MIT); the stored copy get_code serves, from pkwww/SincNet_MLP, is pointer only (licence: NONE) · get_code("b380f42b5b71cc47")
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 pkwww/SincNet_MLP, 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_grads Ran
ran, honoured the contract we drafted
santi-pdp/segan_pytorch/segan/models/ops.py
code served (permissive licence) · get_code("26c162c3caa1223a")
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 1, dtype family float, shape (23,) 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
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        return self.linear2(self.linear1(x))

model = SimpleModel()

# Create some dummy data and perform a forward/backward pass to populate gradients
x = torch.randn(2, 4)
y = torch.randn(2, 2)
criterion = nn.MSELoss()

output = model(x)
loss = criterion(output, y)
loss.backward()

ARGS = {
    "model": model
}
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 1, dtype family float, shape (23,) — shape (23,) 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
l2normalize Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/spectral_norm.py
pointer only (licence: MIT); the stored copy get_code serves, from xxxliu95/RA_FA_Cardiac, is pointer only (licence: NONE) · get_code("bedff51745d2cf84")
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 xxxliu95/RA_FA_Cardiac, 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.

load_config Ran
ran; our contract draft was wrong, not the code
haozh7109/SEGAN-TensorFlow2/SEGAN_TrainMaster.py
code served (permissive licence) · get_code("8058adbabbf8b825")
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 dict. 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 json
import tempfile
import os

# Create a temporary config file with valid JSON content
config_data = {
    "model": "test_model",
    "learning_rate": 0.001,
    "batch_size": 32,
    "epochs": 10,
    "layers": [64, 128, 256],
    "activation": "relu"
}

# Write config to a temporary file
temp_dir = tempfile.mkdtemp()
config_file = os.path.join(temp_dir, "config.json")
with open(config_file, 'w') as f:
    json.dump(config_data, f)

ARGS = {
    "config_file": config_file
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
load_datasets Ran
ran; our contract draft was wrong, not the code
haozh7109/SEGAN-TensorFlow2/Inference_singleShot.py
code served (permissive licence) · get_code("69a97611e4cfefac")
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 2, 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 os
import tempfile
import shutil

# Create temporary directories for testing
temp_dir = tempfile.mkdtemp()
test_clean_dir = os.path.join(temp_dir, 'clean')
test_noisy_dir = os.path.join(temp_dir, 'noisy')

os.makedirs(test_clean_dir, exist_ok=True)
os.makedirs(test_noisy_dir, exist_ok=True)

# Create some dummy files in the test directories
for i in range(3):
    with open(os.path.join(test_clean_dir, f'clean_{i}.txt'), 'w') as f:
        f.write(f'clean file {i}')
    with open(os.path.join(test_noisy_dir, f'noisy_{i}.txt'), 'w') as f:
        f.write(f'noisy file {i}')

# Create a simple config object
class OptConfig:
    def __init__(self, test_clean_data_path, test_noisy_data_path):
        self.test_clean_data_path = test_clean_data_path
        self.test_noisy_data_path = test_noisy_data_path

opt = OptConfig(test_clean_dir, test_noisy_dir)

ARGS = {
    'opt': opt
}
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 tuple, ndim 2, 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
make_divN Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/utils.py
code served (permissive licence) · get_code("cf249ef7bc710ec2")
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 3, dtype family float, shape (2, 6, 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create a small tensor with shape (batch_size=2, sequence_length=5, feature_dim=4)
tensor = torch.randn(2, 5, 4)

# Define parameters
N = 3
method = 'zeros'

# Construct ARGS dict
ARGS = {
    'tensor': tensor,
    'N': N,
    'method': method
}
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 3, dtype family float, shape (2, 6, 4) — shape (2, 6, 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
pos_code Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/core.py
code served (permissive licence) · get_code("2638e76ea8d2f15e")
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 3, dtype family float, shape (2, 4, 3) as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import math

torch.manual_seed(0)

# chunk_pos: a tensor of chunk positions, one per batch element
# x: a tensor of shape (batch_size, pos_dim, chunk_size)
batch_size = 2
pos_dim = 4
chunk_size = 3

chunk_pos = torch.tensor([0, 1], dtype=torch.long)
x = torch.randn(batch_size, pos_dim, chunk_size)

ARGS = {
    "chunk_pos": chunk_pos,
    "x": x,
}
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 3, dtype family float, shape (2, 4, 3) — shape (2, 4, 3) 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
pre_emph Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreser09/segan/data_loader.py
code served (permissive licence) · get_code("4c2a62db536d3e11")
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 EagerTensor, ndim 1, dtype family float, shape (5,) 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 tensorflow as tf
import numpy as np

# Create a small 1D tensor for pre-emphasis
x = tf.constant([0.1, 0.2, 0.3, 0.4, 0.5], dtype=tf.float32)

ARGS = {
    "x": x,
    "coeff": 0.95
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 1, dtype family float, shape (5,) — shape (5,) 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
read_and_slice Ran
ran on a synthesized input
santi-pdp/segan/make_tfrecords.py
code served (permissive licence) · get_code("ab225f7af41a0ef3")
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 ndarray, ndim 2, dtype family int, shape (124, 512) 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 numpy as np
import scipy.io.wavfile as wavfile
import tempfile
import os

# Create a temporary WAV file with 16kHz sampling rate
temp_dir = tempfile.mkdtemp()
temp_wav_path = os.path.join(temp_dir, "test_audio.wav")

# Generate synthetic audio data at 16kHz
sample_rate = 16000
duration = 2  # seconds
n_samples = sample_rate * duration
# Create a simple sine wave
t = np.linspace(0, duration, n_samples)
audio_data = np.sin(2 * np.pi * 440 * t)  # 440 Hz sine wave
# Convert to int16 format (typical for WAV files)
audio_data_int16 = np.int16(audio_data * 32767)

# Write to temporary WAV file
wavfile.write(temp_wav_path, sample_rate, audio_data_int16)

ARGS = {
    "filename": temp_wav_path,
    "wav_canvas_size": 512,
    "stride": 0.5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family int, shape (124, 512) — shape (124, 512) 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
read_and_slice Ran
ran, honoured the contract we drafted
asadullah797/segan_impr/make_tfrecords.py
code served (permissive licence) · get_code("9fd6da88395e5dfe")
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 ndarray, ndim 1, dtype family float, shape (16000,) 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 numpy as np
import scipy.io.wavfile as wavfile
import tempfile
import os

# Create a temporary WAV file for testing
temp_dir = tempfile.mkdtemp()
temp_wav_path = os.path.join(temp_dir, "test_audio.wav")

# Generate a simple 16kHz audio signal (1 second)
sample_rate = 16000
duration = 1  # seconds
num_samples = sample_rate * duration
audio_data = np.random.randn(num_samples).astype(np.float32) * 32767
audio_data = audio_data.astype(np.int16)

# Write the WAV file
wavfile.write(temp_wav_path, sample_rate, audio_data)

# Define ARGS for the function
ARGS = {
    "filename": temp_wav_path,
    "wav_canvas_size": 512,
    "stride": 1.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (16000,) — shape (16000,) 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
scalar_summary Ran
ran on a synthesized input
santi-pdp/segan/ops.py
code served (permissive licence) · get_code("35f13999f3f8f9dd")
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 EagerTensor, 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 tensorflow as tf

# For scalar_summary, we need:
# - name: a string identifying the summary
# - x: a scalar tensor (or something that can be converted to a scalar)

# Create a simple scalar tensor
scalar_tensor = tf.constant(0.5, dtype=tf.float32)

# Define arguments for the function
ARGS = {
    "name": "test_scalar",
    "x": scalar_tensor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
scalar_summary Ran
ran, violated the contract we drafted
andreser09/segan/ops.py
code served (permissive licence) · get_code("610ca41dfb04b88a")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

# Set up TensorFlow to ensure we're in eager execution mode
tf.compat.v1.enable_eager_execution()

# Create a simple scalar tensor
scalar_value = tf.constant(0.5)

# Create a named scalar (can also be a float)
name_str = "test_scalar"

ARGS = {
    "name": name_str,
    "x": scalar_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
sinc Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/models/modules.py
code served (permissive licence) · get_code("7c9569388b69a763")
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 1, dtype family float, shape (11,) as a later run recorded it. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
import math

torch.manual_seed(0)

# The function sinc takes:
# - band: a scalar value used in computation
# - t_right: a tensor used in sinc computation
# - cuda: boolean flag (default False)

# Looking at the function body:
# - band is multiplied with t_right in torch operations
# - t_right is a tensor that gets passed to sin() function
# - cuda is a boolean controlling device placement

# Create a small tensor for t_right (1D tensor with a few values)
t_right = torch.linspace(0.1, 1.0, 5)

# band is a scalar float value
band = 0.5

# cuda should be False (CPU only as per rules)
cuda = False

ARGS = {
    "band": band,
    "t_right": t_right,
    "cuda": cuda
}
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 1, dtype family float, shape (11,) — shape (11,) 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
slice_signal Ran
ran on a synthesized input
santi-pdp/segan/make_tfrecords.py
code served (permissive licence) · get_code("46c1eb78df53cf66")
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 ndarray, ndim 2, dtype family int, shape (7, 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 numpy as np

# Create a simple 1D signal
signal = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], dtype=np.int32)

# Define window_size and stride parameters
window_size = 4
stride = 0.5

ARGS = {
    "signal": signal,
    "window_size": window_size,
    "stride": stride,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family int, shape (7, 4) — shape (7, 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
slice_signal Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
asadullah797/segan_impr/make_tfrecords.py
code served (permissive licence) · get_code("dbce5ca80cee35d6")
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 ndarray, ndim 2, dtype family float, shape (7, 64) 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 numpy as np

# Create a simple 1D signal
signal = np.random.randn(256).astype(np.float32)

# Define window size
window_size = 64

# Define stride as a fraction of window_size
stride = 0.5

ARGS = {
    "signal": signal,
    "window_size": window_size,
    "stride": stride,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (7, 64) — shape (7, 64) 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
slice_signal Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
santi-pdp/segan_pytorch/segan/datasets/se_dataset.py
code served (permissive licence) · get_code("d8d685caed96ccb2")
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 list, ndim 2, 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 numpy as np

# Create a simple 1D signal
signal = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], dtype=np.float32)

# Define window sizes to slice the signal with
window_sizes = [4, 8]

# Define stride as a fraction
stride = 0.5

ARGS = {
    "signal": signal,
    "window_sizes": window_sizes,
    "stride": stride
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family int — a list 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
uttname2spkid Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/utils.py
code served (permissive licence) · get_code("9c7d42585a206ba4")
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 str. 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

def uttname2spkid(uttname):
    spkid = uttname.split('_')[0]
    return spkid

ARGS = {
    'uttname': 'speaker1_utt001_mic1'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
varlen_wav_collate Ran
ran; our contract draft was wrong, not the code
santi-pdp/segan_pytorch/segan/datasets/vc_dataset.py
code served (permissive licence) · get_code("3d5991b1620d158d")
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. 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 numpy as np

torch.manual_seed(0)

# Create sample data for the batch
# Each sample can be either (src, trg) or (id, src, trg)
# src and trg are 1D arrays/tensors representing waveform data

sample1_2elem = (
    np.array([0.1, 0.2, 0.3], dtype=np.float32),  # src
    np.array([0.4, 0.5], dtype=np.float32)  # trg
)

sample2_2elem = (
    np.array([0.6, 0.7, 0.8, 0.9], dtype=np.float32),  # src (longer)
    np.array([0.1, 0.2, 0.3, 0.4, 0.5], dtype=np.float32)  # trg (longer)
)

sample3_3elem = (
    "sample_id_123",  # id
    np.array([0.2, 0.3], dtype=np.float32),  # src
    np.array([0.6], dtype=np.float32)  # trg
)

batch = [sample1_2elem, sample2_2elem, sample3_3elem]

ARGS = {
    "batch": batch
}
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 — 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: none by the harness; any seed is in the setup block
wav_write Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
YosukeSugiura/SEGAN/data.py
pointer only (licence: NONE) · get_code("1b028fcf0c1ec0d5")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, YosukeSugiura/SEGAN 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.

check_input_sampling Not yet run
unverified (harvested, no recorded run)
develooper1994/MasterThesis/clean.py
pointer only (licence: NONE) · get_code("d9f2796539b63330")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, develooper1994/MasterThesis 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. No invocation recorded.

deserialize Not yet run
unverified (harvested, no recorded run)
huylenguyen806/vnasr/tensorflow_asr/callbacks.py
code served (permissive licence) · get_code("efd738a0bc3cc757")
How it ran

no invocation recorded.

downsample Not yet run
unverified (harvested, no recorded run)
develooper1994/MasterThesis/clean.py
pointer only (licence: NONE) · get_code("fba9aaf89912df38")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, develooper1994/MasterThesis 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. No invocation recorded.

gaussian_noise_layer Not yet run
unverified (harvested, no recorded run)
santi-pdp/segan/ops.py
code served (permissive licence) · get_code("7f91d893c058c60b")
How it ran

no invocation recorded.

get_input_sampling Not yet run
unverified (harvested, no recorded run)
develooper1994/MasterThesis/clean.py
pointer only (licence: NONE) · get_code("41f93d7418a0950c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, develooper1994/MasterThesis 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. No invocation recorded.

make_optimizer Not yet run
unverified (harvested, no recorded run)
santi-pdp/segan_pytorch/segan/models/ops.py
code served (permissive licence) · get_code("42bb0a5113fe251c")
How it ran

no invocation recorded.

pre_emph Not yet run
unverified (harvested, no recorded run)
santi-pdp/segan/data_loader.py
code served (permissive licence) · get_code("f00acf17aed0af5a")
How it ran

no invocation recorded.

read_and_decode Not yet run
unverified (harvested, no recorded run)
santi-pdp/segan/data_loader.py
code served (permissive licence) · get_code("28105186fd5104d4")
How it ran

no invocation recorded.

read_and_decode Not yet run
unverified (harvested, no recorded run)
andreser09/segan/data_loader.py
code served (permissive licence) · get_code("534a5aaa629f0b62")
How it ran

no invocation recorded.

read_and_decode Not yet run
unverified (harvested, no recorded run)
asadullah797/segan_impr/data_loader.py
code served (permissive licence) · get_code("455f04cbcd882c4d")
How it ran

no invocation recorded.

read_and_decode Not yet run
unverified (harvested, no recorded run)
ssarfjoo/improvedsegan/data_loader.py
code served (permissive licence) · get_code("67b26950d2b2b2b8")
How it ran

no invocation recorded.

read_and_decode Not yet run
unverified (harvested, no recorded run)
sufengniu/segan_multi/data_loader.py
code served (permissive licence) · get_code("c4f1ac5600121aba")
How it ran

no invocation recorded.

sample_random_walk Not yet run
unverified (harvested, no recorded run)
santi-pdp/segan/ops.py
code served (permissive licence) · get_code("b6fb2cc68fa10225")
How it ran

no invocation recorded.

Also present in other papers' code

Of 39 samples lifted from this paper's linked repositories, 39 carry a function identity and 0 do not. Of the 38 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 35 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
l2normalize 27 other papers
27 later · 2 only as an assembly brick
1802.05957 (2018-02)
Spectral Normalization for Generative Adversarial Networks
1806.03589 (2018-06)
Free-Form Image Inpainting with Gated Convolution
1810.08771 (2018-10)
Image Inpainting via Generative Multi-column Convolutional Neural Networks
and 24 more not named here
flip 4 other papers
4 later
1808.00158 (2018-08)
Speaker Recognition from Raw Waveform with SincNet
1810.12735 (2018-10)
Spoken Language Understanding on the Edge
1910.09463 (2019-10)
Using Speech Synthesis to Train End-to-End Spoken Language Understanding Models
and 1 more not named here
deserialize 1 other paper
1 later
2005.08100 (2020-05)
Conformer: Convolution-augmented Transformer for Speech Recognition

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

Current speech enhancement techniques operate on the spectral domain and/or exploit some higher-level feature. The majority of them tackle a limited number of noise conditions and rely on first-order statistics. To circumvent these issues, deep networks are being increasingly used, thanks to their ability to learn complex functions from large example sets. In this work, we propose the use of generative adversarial networks for speech enhancement. In contrast to current techniques, we operate at the waveform level, training the model end-to-end, and incorporate 28 speakers and 40 different noise conditions into the same model, such that model parameters are shared across them. We evaluate the proposed model using an independent, unseen test set with two speakers and 20 alternative noise conditions. The enhanced samples confirm the viability of the proposed model, and both objective and subjective evaluations confirm the effectiveness of it. With that, we open the exploration of generative architectures for speech enhancement, which may progressively incorporate further speech-centric design choices to improve their performance.

For agents

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

get_harvested_code_for_paper("1703.09452")
get_code_for_paper("1703.09452")
have("1703.09452")

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

curl https://syntology.ai/api/ran/1703.09452.json

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

[![Syntology run record](https://syntology.ai/api/ran/1703.09452.svg)](https://syntology.ai/paper/1703.09452)

Connect an agent — have() is free.