SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2110.03370 · 2021

WenetSpeech: A 10000+ Hours Multi-domain Mandarin Corpus for Speech Recognition

arXiv · PDF · Open in the Atlas

Code that ran

8 of 8 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 8 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/2110.03370.json · badge: /api/ran/2110.03370.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
wenet-e2e/wenetspeech canonical (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
aizhiqi-work/MM-KWS community (classifier label)
classifier basis not recorded on the link
7 ran · 0 no recorded run
FunctionStatusWhere it lives
collate_fn Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/SPC_N0_ALL.py
code served (permissive licence) · get_code("896713b2dccaa1db")
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
from torch.nn.utils.rnn import pad_sequence

torch.manual_seed(0)

# Create sample batch items matching the unpacking pattern:
# ids, fbank_feature, g2p_embed, lm_embed, Query_label, Support_label, label
batch_item_1 = (
    "id_1",  # ids
    torch.randn(5, 4),  # fbank_feature: sequence length 5, feature dim 4
    torch.randn(5, 3),  # g2p_embed: sequence length 5, embedding dim 3
    torch.randn(1, 2),  # lm_embed: 1 x 2
    "query_1",  # Query_label
    "support_1",  # Support_label
    0  # label
)

batch_item_2 = (
    "id_2",  # ids
    torch.randn(7, 4),  # fbank_feature: sequence length 7, feature dim 4
    torch.randn(7, 3),  # g2p_embed: sequence length 7, embedding dim 3
    torch.randn(1, 2),  # lm_embed: 1 x 2
    "query_2",  # Query_label
    "support_2",  # Support_label
    1  # label
)

batch = [batch_item_1, batch_item_2]

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 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
collate_fn Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/SPC_N1_ALL.py
code served (permissive licence) · get_code("0bf83f224a6bea55")
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
from torch.nn.utils.rnn import pad_sequence

torch.manual_seed(0)

# Create sample batch items
# Each item in batch: (ids, fbank_feature, g2p_embed, lm_embed, audiolm_embed, Query_label, Support_label, label)

batch_item_1 = (
    "id_1",  # ids
    torch.randn(5, 4),  # fbank_feature: sequence length 5, feature dim 4
    torch.randn(6, 3),  # g2p_embed: sequence length 6, embedding dim 3
    torch.randn(1, 2),  # lm_embed: 1 x 2
    torch.randn(1, 2),  # audiolm_embed: 1 x 2
    torch.tensor([1, 0, 1]),  # Query_label
    torch.tensor([0, 1]),  # Support_label
    1  # label
)

batch_item_2 = (
    "id_2",  # ids
    torch.randn(4, 4),  # fbank_feature: sequence length 4, feature dim 4
    torch.randn(5, 3),  # g2p_embed: sequence length 5, embedding dim 3
    torch.randn(1, 2),  # lm_embed: 1 x 2
    torch.randn(1, 2),  # audiolm_embed: 1 x 2
    torch.tensor([0, 1, 0]),  # Query_label
    torch.tensor([1, 0]),  # Support_label
    0  # label
)

batch = [batch_item_1, batch_item_2]

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 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
collate_fn Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/libriphrase_test_18.py
code served (permissive licence) · get_code("1869bcafdc7df027")
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
from torch.nn.utils.rnn import pad_sequence

torch.manual_seed(0)

# Create sample batch items matching the unpacking: 
# fbank_feature, g2p_embed, lm_embed, audiolm_embed, label
batch_item_1 = (
    torch.randn(5, 4),      # fbank_feature: sequence of length 5, feature dim 4
    torch.randn(3, 4),      # g2p_embed: sequence of length 3, embedding dim 4
    torch.randn(2, 4),      # lm_embed: sequence of length 2, embedding dim 4
    torch.randn(1, 4),      # audiolm_embed: sequence of length 1, embedding dim 4
    0                       # label: class index
)

batch_item_2 = (
    torch.randn(4, 4),      # fbank_feature: sequence of length 4, feature dim 4
    torch.randn(2, 4),      # g2p_embed: sequence of length 2, embedding dim 4
    torch.randn(3, 4),      # lm_embed: sequence of length 3, embedding dim 4
    torch.randn(1, 4),      # audiolm_embed: sequence of length 1, embedding dim 4
    1                       # label: class index
)

batch = [batch_item_1, batch_item_2]

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 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
get_files Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/wenetphrase_train.py
code served (permissive licence) · get_code("1d9a26a31a50681a")
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 1, 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 a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()

# Create subdirectories and files
os.makedirs(os.path.join(temp_dir, 'subdir1'), exist_ok=True)
os.makedirs(os.path.join(temp_dir, 'subdir2'), exist_ok=True)

# Create some .wav files
open(os.path.join(temp_dir, 'file1.wav'), 'w').close()
open(os.path.join(temp_dir, 'subdir1', 'file2.wav'), 'w').close()
open(os.path.join(temp_dir, 'subdir1', 'file3.mp3'), 'w').close()
open(os.path.join(temp_dir, 'subdir2', 'file4.wav'), 'w').close()
open(os.path.join(temp_dir, 'file5.txt'), 'w').close()

ARGS = {
    'path': temp_dir,
    'endswith': ['.wav']
}
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 list, ndim 1, 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
get_files Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/SPC_N0_ALL.py
code served (permissive licence) · get_code("7f08f754c8e83bee")
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 1, 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 a temporary directory structure for testing
temp_dir = tempfile.mkdtemp()

# Create some test files with different extensions
os.makedirs(os.path.join(temp_dir, "subdir1"), exist_ok=True)
os.makedirs(os.path.join(temp_dir, "subdir2"), exist_ok=True)

# Create test files
open(os.path.join(temp_dir, "file1.txt"), "w").close()
open(os.path.join(temp_dir, "file2.py"), "w").close()
open(os.path.join(temp_dir, "subdir1", "file3.txt"), "w").close()
open(os.path.join(temp_dir, "subdir1", "file4.py"), "w").close()
open(os.path.join(temp_dir, "subdir2", "file5.txt"), "w").close()

ARGS = {
    "path": temp_dir,
    "endswith": ".txt"
}
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 list, ndim 1, 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
padding_and_mask Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/wenetphrase_train.py
code served (permissive licence) · get_code("3b8a0af5bbc57b93")
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 3, 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 expects 'data' to be an iterable of sequences
# Each sequence should support len() and be iterable
# The function pads sequences and creates masks

# Create a list of sequences with varying lengths
sequence1 = [1, 2, 3]
sequence2 = [4, 5]
sequence3 = [6, 7, 8, 9]

data = [sequence1, sequence2, sequence3]

ARGS = {
    "data": data
}
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 3, 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
read_file Ran
ran on a synthesized input
wenet-e2e/wenetspeech/toolkits/wenet/local/process_opus.py
code served (permissive licence) · get_code("b7cc041a8cc401d7")
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 tempfile
import os

# Create temporary files for wav_scp and segments
temp_dir = tempfile.mkdtemp()
wav_scp_file = os.path.join(temp_dir, "wav.scp")
segments_file = os.path.join(temp_dir, "segments")

# Write wav.scp file with format: wav_id path
with open(wav_scp_file, 'w', encoding='UTF-8') as f:
    f.write("wav_001 /path/to/audio1.wav\n")
    f.write("wav_002 /path/to/audio2.wav\n")
    f.write("wav_003 /path/to/audio3.wav\n")

# Write segments file with format: utt_id wav_id start_time end_time
with open(segments_file, 'w', encoding='UTF-8') as f:
    f.write("utt_001 wav_001 0.0 1.5\n")
    f.write("utt_002 wav_001 1.5 3.2\n")
    f.write("utt_003 wav_002 0.5 2.1\n")

ARGS = {
    "wav_scp": wav_scp_file,
    "segments": segments_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 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
train_collate_fn Ran
ran on a synthesized input
aizhiqi-work/MM-KWS/mm-kws/dataloaders/wenetphrase_train.py
code served (permissive licence) · get_code("0d74ccb2ff7f766a")
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 small sample data matching the expected structure
# batch is a list of tuples: (fbank_feature, g2p_embed, lm_embed, audiolm_embed, label, Phoneme_label, Text_label)

batch_item_1 = (
    # fbank_feature: list of tensors
    [torch.randn(5, 4), torch.randn(3, 4)],
    # g2p_embed: list of numpy arrays
    [np.random.randn(2, 4), np.random.randn(3, 4)],
    # lm_embed: list of numpy arrays (with extra dimension)
    [np.random.randn(2, 1, 4), np.random.randn(3, 1, 4)],
    # audiolm_embed: list of tensors
    [torch.randn(2, 4), torch.randn(3, 4)],
    # label: list of ints
    [0, 1],
    # Phoneme_label: list of lists of ints
    [[1, 2, 3], [4, 5]],
    # Text_label: list of lists of ints
    [[10, 11], [12, 13, 14]],
)

batch_item_2 = (
    # fbank_feature: list of tensors
    [torch.randn(4, 4), torch.randn(2, 4)],
    # g2p_embed: list of numpy arrays
    [np.random.randn(2, 4), np.random.randn(2, 4)],
    # lm_embed: list of numpy arrays (with extra dimension)
    [np.random.randn(2, 1, 4), np.random.randn(2, 1, 4)],
    # audiolm_embed: list of tensors
    [torch.randn(2, 4), torch.randn(2, 4)],
    # label: list of ints
    [1, 0],
    # Phoneme_label: list of lists of ints
    [[2, 3], [5, 6, 7]],
    # Text_label: list of lists of ints
    [[11, 12, 13], [14, 15]],
)

batch = [batch_item_1, batch_item_2]

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

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 8 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.

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

In this paper, we present WenetSpeech, a multi-domain Mandarin corpus consisting of 10000+ hours high-quality labeled speech, 2400+ hours weakly labeled speech, and about 10000 hours unlabeled speech, with 22400+ hours in total. We collect the data from YouTube and Podcast, which covers a variety of speaking styles, scenarios, domains, topics, and noisy conditions. An optical character recognition (OCR) based method is introduced to generate the audio/text segmentation candidates for the YouTube data on its corresponding video captions, while a high-quality ASR transcription system is used to generate audio/text pair candidates for the Podcast data. Then we propose a novel end-to-end label error detection approach to further validate and filter the candidates. We also provide three manually labelled high-quality test sets along with WenetSpeech for evaluation -- Dev for cross-validation purpose in training, Test_Net, collected from Internet for matched test, and Test\_Meeting, recorded from real meetings for more challenging mismatched test. Baseline systems trained with WenetSpeech are provided for three popular speech recognition toolkits, namely Kaldi, ESPnet, and WeNet, and recognition results on the three test sets are also provided as benchmarks. To the best of our knowledge, WenetSpeech is the current largest open-sourced Mandarin speech corpus with transcriptions, which benefits research on production-level speech recognition.

For agents

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

get_harvested_code_for_paper("2110.03370")
get_code_for_paper("2110.03370")
have("2110.03370")

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

curl https://syntology.ai/api/ran/2110.03370.json

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

[![Syntology run record](https://syntology.ai/api/ran/2110.03370.svg)](https://syntology.ai/paper/2110.03370)

Connect an agent — have() is free.