SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1803.05337 · 2018

Learning to Recognize Musical Genre from Audio

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 1 violated it and 7 have no contract verdict; 1 failure of our instrument; 3 no recorded run); 1 of them was also run on the host, not in the jail, by the peer-agreement lane that set its level, 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/1803.05337.json · badge: /api/ran/1803.05337.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
crowdAI/crowdai-musical-genre-recognition-starter-kit canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
raonsol/deep-pitcher community (classifier label)
classifier basis not recorded on the link
4 ran · 2 no recorded run
bhavnasud/MusicRecognition community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
softmax Ran
ran, violated the contract we drafted
crowdAI/crowdai-musical-genre-recognition-starter-kit/random_submission.py
code served (permissive licence) · get_code("7f9b15275ca5759c")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-05T00:37:48Z). 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

# The softmax function takes a single parameter 'x' which should be a numpy array
# It performs exponential normalization on the input

# Create a small test input - a 1D array
x = np.array([1.0, 2.0, 3.0, 4.0])

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
build_sample_loader Ran
ran on a synthesized input
bhavnasud/MusicRecognition/utils.py
code served (permissive licence) · get_code("1d941b22c0892036")
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 type. 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 numpy as np
import pandas as pd
import os.path

# Create a minimal loader class that has the required 'shape' attribute and 'load' method
class MinimalAudioLoader:
    def __init__(self):
        self.shape = (128,)  # Audio feature shape
    
    def load(self, path):
        # Return dummy audio features of the correct shape
        return np.random.randn(*self.shape).astype(np.float32)

# Create a minimal Y dataframe with track_id as index and some labels
Y = pd.DataFrame(
    np.random.randint(0, 5, size=(10, 3)),  # 10 tracks, 3 label columns
    index=np.arange(10),
    columns=['label_0', 'label_1', 'label_2']
)
Y.index.name = 'track_id'

# Create a minimal audio directory structure
audio_dir = '/tmp/audio_data'

# Create the loader instance
loader = MinimalAudioLoader()

# Create track IDs array for the SampleLoader
tids = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9], dtype=np.int32)

ARGS = {
    'audio_dir': audio_dir,
    'Y': Y,
    'loader': loader
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
convert_duration Ran
ran, honoured the contract we drafted
bhavnasud/MusicRecognition/creation.py
code served (permissive licence) · get_code("46a60ba365931494")
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 torch
import numpy as np

def convert_duration(x):
    times = x.split(':')
    seconds = int(times[-1])
    minutes = int(times[-2])
    try:
        minutes += 60 * int(times[-3])
    except IndexError:
        pass
    return seconds + 60 * minutes

# The function takes a string parameter 'x' that represents time in HH:MM:SS or MM:SS format
# It splits by ':' and converts to integers, then calculates total seconds

ARGS = {
    'x': '1:30:45'  # Format: HH:MM:SS (1 hour, 30 minutes, 45 seconds)
}
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 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
get_audio_path Ran
ran; our contract draft was wrong, not the code
bhavnasud/MusicRecognition/utils.py
code served (permissive licence) · get_code("f7bb5dc60935c4cc")
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 os.path

# Construct arguments for get_audio_path
# The function takes:
# 1. audio_dir: a directory path (string)
# 2. track_id: an integer track identifier

# Looking at the function body:
# - It formats track_id as a 6-digit zero-padded string
# - It extracts the first 3 characters as a subdirectory
# - It constructs a path using os.path.join

# Create a valid audio directory path
audio_dir = '../data/fma_small'

# Create a valid track_id (integer)
track_id = 2

ARGS = {
    'audio_dir': audio_dir,
    'track_id': track_id
}
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 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
get_audio_path Ran
ran on a synthesized input
crowdAI/crowdai-musical-genre-recognition-starter-kit/fma.py
code served (permissive licence) · get_code("bd3fdd3075af65cd")
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 os

def get_audio_path(track_id):
    tid_str = '{:06d}'.format(track_id)
    return os.path.join('data', 'fma_medium', tid_str[:3], tid_str + '.mp3')

ARGS = {
    'track_id': 42
}
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 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
get_columns Ran
ran on a synthesized input
raonsol/deep-pitcher/features.py
code served (permissive licence) · get_code("0fce03faa5ea512e")
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 MultiIndex. 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 pandas as pd

# The function expects:
# - feature_list: a dict mapping feature names (str) to sizes (int)
# - moments: an iterable of moment names (str)
# - prefix: optional string prefix (can be None)

feature_list = {
    'mfcc': 13,
    'chroma': 12,
    'tempo': 1,
    'spectral_centroid': 5
}

moments = ['mean', 'std', 'min', 'max']

prefix = 'audio'

ARGS = {
    'feature_list': feature_list,
    'moments': moments,
    'prefix': prefix
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MultiIndex — a MultiIndex output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MultiIndex output the recorded ndim 1 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
get_filename Ran
ran on a synthesized input
raonsol/deep-pitcher/utils.py
code served (permissive licence) · get_code("c740b4b2518c3dac")
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 os

# The function get_filename takes a path (String or PathLike object)
# and returns the filename without extension using os.path operations.
# It doesn't actually read from the filesystem, just manipulates path strings.

in_path = "/home/user/documents/example_file.txt"

ARGS = {
    "in_path": in_path
}
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 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
get_paths Ran
ran on a synthesized input
raonsol/deep-pitcher/utils.py
code served (permissive licence) · get_code("5150c03fa3be092b")
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

# Create a temporary directory with some test files
temp_dir = tempfile.mkdtemp()

# Create some test files in the temporary directory
test_files = ['file1.txt', 'file2.py', 'file3.csv']
for filename in test_files:
    filepath = os.path.join(temp_dir, filename)
    with open(filepath, 'w') as f:
        f.write('test content')

# Also create a hidden file that should be excluded
hidden_file = os.path.join(temp_dir, '.hidden')
with open(hidden_file, 'w') as f:
    f.write('hidden')

# Create a subdirectory that should be excluded
subdir = os.path.join(temp_dir, 'subdir')
os.makedirs(subdir)

ARGS = {
    'in_path': temp_dir
}
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
load Ran
ran on a synthesized input
crowdAI/crowdai-musical-genre-recognition-starter-kit/fma.py
code served (permissive licence) · get_code("72f77230f9b719e5")
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 DataFrame. 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 os
import tempfile
import pandas as pd

# Create temporary CSV files with appropriate structure for each case

# Create a temporary directory
temp_dir = tempfile.mkdtemp()

# 1. features file with multi-level columns (0, 1, 2)
features_path = os.path.join(temp_dir, "features.csv")
features_df = pd.DataFrame(
    [[1.0, 2.0], [3.0, 4.0]],
    index=['song1', 'song2'],
    columns=pd.MultiIndex.from_tuples([('level0_a', 'level1_a', 'level2_a'), ('level0_b', 'level1_b', 'level2_b')])
)
features_df.index.name = 'track_id'
features_df.to_csv(features_path)

# 2. echonest file with multi-level columns (0, 1, 2)
echonest_path = os.path.join(temp_dir, "echonest.csv")
echonest_df = pd.DataFrame(
    [[5.0, 6.0], [7.0, 8.0]],
    index=['track1', 'track2'],
    columns=pd.MultiIndex.from_tuples([('echo_level0_a', 'echo_level1_a', 'echo_level2_a'), ('echo_level0_b', 'echo_level1_b', 'echo_level2_b')])
)
echonest_df.index.name = 'track_id'
echonest_df.to_csv(echonest_path)

# 3. genres file with single level index
genres_path = os.path.join(temp_dir, "genres.csv")
genres_df = pd.DataFrame(
    {'genre_name': ['rock', 'pop']},
    index=['genre1', 'genre2']
)
genres_df.index.name = 'genre_id'
genres_df.to_csv(genres_path)

# 4. tracks file with multi-level columns (0, 1) and complex data
tracks_path = os.path.join(temp_dir, "tracks.csv")
tracks_data = {
    ('track', 'tags'): ["['tag1', 'tag2']", "['tag3']"],
    ('album', 'tags'): ["['album_tag1']", "['album_tag2', 'album_tag3']"],
    ('artist', 'tags'): ["['artist_tag1', 'artist_tag2']", "['artist_tag3']"],
    ('track', 'genres'): ["[1, 2]", "[3]"],
    ('track', 'genres_all'): ["[1, 2, 4]", "[3, 5]"],
    ('track', 'date_created'): ['2020-01-01', '2020-06-15'],
    ('track', 'date_recorded'): ['2019-01-01', '2019-06-15'],
    ('album', 'date_created'): ['2018-01-01', '2018-06-15'],
    ('album', 'date_released'): ['2018-02-01', '2018-07-15'],
    ('artist', 'date_created'): ['2017-01-01', '2017-06-15'],
    ('artist', 'active_year_begin'): ['2000-01-01', '2005-01-01'],
    ('artist', 'active_year_end'): ['2020-01-01', '2021-01-01'],
    ('track', 'genre_top'): ['rock', 'pop'],
    ('track', 'license'): ['cc-by', 'cc-by-sa'],
    ('album', 'type'): ['album', 'ep'],
    ('album', 'information'): ['info1', 'info2'],
    ('artist', 'bio'): ['bio1', 'bio2'],
    ('set', 'subset'): ['small', 'medium'],
}
tracks_df = pd.DataFrame(tracks_data, index=['track1', 'track2'])
tracks_df.columns = pd.MultiIndex.from_tuples(tracks_df.columns)
tracks_df.index.name = 'track_id'
tracks_df.to_csv(tracks_path)

ARGS = {
    'filepath': features_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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_xml Ran
ran on a synthesized input
raonsol/deep-pitcher/itdbloader.py
code served (permissive licence) · get_code("e94b273a39fdef9f")
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 tempfile
import os
import plistlib
from pathlib import Path

# Create a temporary XML/plist file for testing
temp_dir = tempfile.mkdtemp()
test_plist_path = os.path.join(temp_dir, "test.plist")

# Create a simple plist file with test data
test_data = {
    "key1": "value1",
    "key2": 42,
    "key3": [1, 2, 3],
    "key4": {"nested": "dict"}
}

with open(test_plist_path, 'wb') as f:
    plistlib.dump(test_data, f)

ARGS = {
    "xmlfile": test_plist_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
humanize — 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 Not yet run
unverified (harvested, no recorded run)
bhavnasud/MusicRecognition/utils.py
code served (permissive licence) · get_code("3decdbb26280ee7e")
How it ran

no invocation recorded.

send_patch Not yet run
unverified (harvested, no recorded run)
raonsol/deep-pitcher/import_DB.py
code served (permissive licence) · get_code("bed3cace90df5757")
How it ran

no invocation recorded.

yes_or_no Not yet run
unverified (harvested, no recorded run)
raonsol/deep-pitcher/utils.py
code served (permissive licence) · get_code("3a6d837b871493f2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 3 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 3 later) and 10 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
softmax 12 other papers
1 dated earlier, 11 later
1411.2738 (2014-11)
word2vec Parameter Learning Explained
1906.05394 (2019-06)
Neural Arabic Question Answering
2103.15692 (2021-03)
Self-Constructing Neural Networks Through Random Mutation
and 9 more not named here
convert_duration 1 other paper
1 later
2310.07160 (2023-10)
A Multimodal Instruction-Following Language Model for Music
get_audio_path 1 other paper
1 later
2310.07160 (2023-10)
A Multimodal Instruction-Following Language Model for Music

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

Repositories linked to this paper

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

Abstract

We here summarize our experience running a challenge with open data for musical genre recognition. Those notes motivate the task and the challenge design, show some statistics about the submissions, and present the results.

For agents

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

get_harvested_code_for_paper("1803.05337")
get_code_for_paper("1803.05337")
have("1803.05337")

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

curl https://syntology.ai/api/ran/1803.05337.json

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

[![Syntology run record](https://syntology.ai/api/ran/1803.05337.svg)](https://syntology.ai/paper/1803.05337)

Connect an agent — have() is free.