SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.13995 · 2020

One-class Learning Towards Synthetic Voice Spoofing Detection

arXiv · PDF · Open in the Atlas

Code that ran

10 of 10 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; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 0 failures of our instrument; 0 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/2010.13995.json · badge: /api/ran/2010.13995.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
nii-yamagishilab/project-NN-Pytorch-scripts community (classifier label)
classifier basis not recorded on the link
10 ran · 0 no recorded run
FunctionStatusWhere it lives
GriffinLim Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/dsp_tools.py
code served (permissive licence) · get_code("b683db46244e0bc0")
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 (2304,) 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

# Set seed for reproducibility
np.random.seed(0)

# Create synthetic spectrum amplitude data
# Shape should be (fft_n//2+1, n_frames) or (n_frames, fft_n//2+1)
fft_n = 512
n_frames = 10
n_freqs = fft_n // 2 + 1

# Create realistic spectrum amplitude (linear domain, non-negative)
sp_amp = np.abs(np.random.randn(n_freqs, n_frames)).astype(np.float32) + 0.1

# Parameters for STFT
fl = 512  # frame length
fs = 256  # frame shift
n_iter = 3  # number of Griffin-Lim iterations

ARGS = {
    'sp_amp': sp_amp,
    'n_iter': n_iter,
    'fl': fl,
    'fs': fs,
    'fft_n': fft_n,
    'window': 'hann',
    'momentum': 0.99,
    'init': 'rand'
}
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 (2304,) — shape (2304,) 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
customize_collate Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/customize_collate_fn.py
code served (permissive licence) · get_code("80ea715b2cca2a19")
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 (3, 5, 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 collections

torch.manual_seed(0)

# Define string_classes for compatibility
try:
    string_classes = (str,)
except:
    string_classes = (str,)

# Create a batch of tensors with different lengths
# Each tensor has shape (seq_len, dim) where seq_len varies
tensor1 = torch.randn(3, 4)  # sequence length 3, dimension 4
tensor2 = torch.randn(5, 4)  # sequence length 5, dimension 4
tensor3 = torch.randn(2, 4)  # sequence length 2, dimension 4

batch = [tensor1, tensor2, tensor3]

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 Tensor, ndim 3, dtype family float, shape (3, 5, 4) — shape (3, 5, 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
f0resize Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/dsp_tools.py
code served (permissive licence) · get_code("34061140db57883e")
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 (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 f0 array with some frequency values and some zeros (unvoiced frames)
input_f0 = np.array([0, 100.5, 150.2, 0, 200.1, 180.3, 0, 0], dtype=np.float32)

# Input resolution: 10ms frame shift
input_reso = 10

# Output resolution: 20ms frame shift (upsampling)
output_reso = 20

ARGS = {
    "input_f0": input_f0,
    "input_reso": input_reso,
    "output_reso": output_reso,
}
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 (4,) — shape (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
f_args_parsed Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/config_parse/arg_parse.py
code served (permissive licence) · get_code("8ebf959aff5951c9")
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 Namespace. 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 argparse

torch.manual_seed(0)

ARGS = {
    'argument_input': None
}
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 Namespace — a Namespace output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Namespace 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
f_read_htk Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/io_tools.py
code served (permissive licence) · get_code("78c42439140c4715")
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 (10, 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 tempfile
import os

# Create a temporary HTK file for testing
temp_dir = tempfile.gettempdir()
temp_file = os.path.join(temp_dir, 'test_htk_file.htk')

# Write a valid HTK file with little endian format
with open(temp_file, 'wb') as f:
    # Header: nSample (int32), Period (int32), SampleSize (int16), kind (int16)
    # Using little endian format
    nSample = np.int32(10)  # 10 samples
    Period = np.int32(100000)  # 100ms in 100ns units
    SampleSize = np.int16(16)  # 16 bytes = 4 floats of 4 bytes each
    kind = np.int16(6)  # MFCC
    
    f.write(nSample.tobytes())
    f.write(Period.tobytes())
    f.write(SampleSize.tobytes())
    f.write(kind.tobytes())
    
    # Write data: 10 samples, each with 4 floats
    for i in range(10):
        sample = np.array([1.0, 2.0, 3.0, 4.0], dtype=np.float32)
        f.write(sample.tobytes())

ARGS = {
    'filename': temp_file,
    'data_format': 'f4',
    'end': 'l'
}
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 (10, 4) — shape (10, 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
f_read_raw_mat Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/io_tools.py
code served (permissive licence) · get_code("44f348fd3f9d6e71")
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 (5, 3) 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 tempfile
import os

# Create a temporary binary file with test data
temp_dir = tempfile.mkdtemp()
filename = os.path.join(temp_dir, 'test_data.bin')

# Create sample binary data: 5 rows × 3 columns of float32 values
col = 3
num_rows = 5
data = np.arange(num_rows * col, dtype=np.float32).reshape(num_rows, col)

# Write as little-endian binary
with open(filename, 'wb') as f:
    data.astype('<f4').tofile(f)

ARGS = {
    'filename': filename,
    'col': col,
    'data_format': 'f4',
    'end': 'l'
}
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 (5, 3) — shape (5, 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
f_read_raw_mat_length Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/io_tools.py
code served (permissive licence) · get_code("4a01fffa6219c3b2")
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 int, ndim 0, dtype family int, 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 tempfile
import struct
import os

# Create a temporary binary file with known content
temp_file = tempfile.NamedTemporaryFile(mode='wb', delete=False, suffix='.bin')
temp_filename = temp_file.name

# Write some float32 data (4 bytes per float)
# Writing 10 float32 values = 40 bytes total
num_elements = 10
for i in range(num_elements):
    temp_file.write(struct.pack('f', float(i)))

temp_file.close()

ARGS = {
    'filename': temp_filename,
    'data_format': 'f4'
}
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 int, ndim 0, dtype family int, 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
pad_sequence Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/customize_collate_fn.py
code served (permissive licence) · get_code("3cbf08a7a33e95af")
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 3, 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
import torch.nn as nn

torch.manual_seed(0)

# Create a batch of tensors with different sequence lengths but same trailing dimensions
# Each tensor has shape (sequence_length, feature_dim)
tensor1 = torch.randn(5, 4)  # sequence length 5, feature dim 4
tensor2 = torch.randn(3, 4)  # sequence length 3, feature dim 4
tensor3 = torch.randn(7, 4)  # sequence length 7, feature dim 4

batch = [tensor1, tensor2, tensor3]

ARGS = {
    'batch': batch,
    'padding_value': 0.0
}
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, ndim 3, dtype family float — 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
pad_sequence_batch Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/customize_collate_fn.py
code served (permissive licence) · get_code("90eb3647bd4bd702")
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 5, 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
import torch.nn as nn

torch.manual_seed(0)

# Construct list_batch: a list of batched tensors with different sequence lengths
# Each tensor has shape (batch_size, seq_len, dim1, dim2, ...)
# We'll create 3 batches with different sequence lengths but same trailing dimensions

batch_1 = torch.randn(2, 3, 4, 5)  # batch_size=2, seq_len=3, trailing_dims=(4, 5)
batch_2 = torch.randn(2, 5, 4, 5)  # batch_size=2, seq_len=5, trailing_dims=(4, 5)
batch_3 = torch.randn(2, 4, 4, 5)  # batch_size=2, seq_len=4, trailing_dims=(4, 5)

list_batch = [batch_1, batch_2, batch_3]

ARGS = {
    "list_batch": list_batch,
    "padding_value": 0.0
}
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, ndim 5, dtype family float — 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
spectra_substraction Ran
ran on a synthesized input
nii-yamagishilab/project-NN-Pytorch-scripts/core_scripts/data_io/dsp_tools.py
code served (permissive licence) · get_code("f08f88c4f095cf72")
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 (4096,) 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

np.random.seed(0)

# Create simple waveforms for denoising
# input1: waveform to be denoised
signal_length1 = 4096
input1 = np.random.randn(signal_length1).astype(np.float32)

# input2: background noise waveform
signal_length2 = 2048
input2 = np.random.randn(signal_length2).astype(np.float32) * 0.1

ARGS = {
    'input1': input1,
    'input2': input2,
    'ratio': 0.1,
    'frame_length': 512,
    'frame_shift': 256,
    'fft_n': 512
}
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 (4096,) — shape (4096,) 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

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 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 7 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
f_read_htk 1 other paper
1 later
2111.04433 (2021-11)
RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofing
f_read_raw_mat 1 other paper
1 later
2111.04433 (2021-11)
RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofing
f_read_raw_mat_length 1 other paper
1 later
2111.04433 (2021-11)
RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofing

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

Human voices can be used to authenticate the identity of the speaker, but the automatic speaker verification (ASV) systems are vulnerable to voice spoofing attacks, such as impersonation, replay, text-to-speech, and voice conversion. Recently, researchers developed anti-spoofing techniques to improve the reliability of ASV systems against spoofing attacks. However, most methods encounter difficulties in detecting unknown attacks in practical use, which often have different statistical distributions from known attacks. Especially, the fast development of synthetic voice spoofing algorithms is generating increasingly powerful attacks, putting the ASV systems at risk of unseen attacks. In this work, we propose an anti-spoofing system to detect unknown synthetic voice spoofing attacks (i.e., text-to-speech or voice conversion) using one-class learning. The key idea is to compact the bona fide speech representation and inject an angular margin to separate the spoofing attacks in the embedding space. Without resorting to any data augmentation methods, our proposed system achieves an equal error rate (EER) of 2.19% on the evaluation set of ASVspoof 2019 Challenge logical access scenario, outperforming all existing single systems (i.e., those without model ensemble).

For agents

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

get_harvested_code_for_paper("2010.13995")
get_code_for_paper("2010.13995")
have("2010.13995")

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

curl https://syntology.ai/api/ran/2010.13995.json

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

[![Syntology run record](https://syntology.ai/api/ran/2010.13995.svg)](https://syntology.ai/paper/2010.13995)

Connect an agent — have() is free.