SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.09079 · 2022

Music Source Separation with Generative Flow

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 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 4 have no contract verdict; 1 failure 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/2204.09079.json · badge: /api/ran/2204.09079.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
gzhu06/generativesourceseparation canonical (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
FunctionStatusWhere it lives
get_spec Ran
ran on a synthesized input
gzhu06/generativesourceseparation/inverse_utils.py
code served (permissive licence) · get_code("d630d9581143f970")
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 a small stand-in STFT class that has the stft_fn attribute
# with a transform method that returns magnitude and phase
class StandInSTFT:
    def __init__(self):
        self.stft_fn = self
    
    def transform(self, audio_norm):
        # audio_norm is expected to be a 1D tensor
        # Return magnitude (with batch dimension) and phase
        batch_size = 1
        freq_bins = 4
        time_steps = 3
        
        # magnitude shape: (batch, freq, time)
        mag = torch.randn(batch_size, freq_bins, time_steps)
        # phase shape: (batch, freq, time) or similar
        pha = torch.randn(batch_size, freq_bins, time_steps)
        
        return mag, pha

# Create audio as numpy array (small size)
audio = np.random.randn(64).astype(np.float32) * 16000

# Create the stft object
stft = StandInSTFT()

ARGS = {
    "audio": audio,
    "stft": stft
}
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
intersperse Ran
ran; our contract draft was wrong, not the code
gzhu06/generativesourceseparation/generator/glow/commons.py
pointer only (licence: MIT); the stored copy get_code serves, from wxzyd123/pivotal_objective_perturbation, is pointer only (licence: NONE) · get_code("c01a3efb274ff778")
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 wxzyd123/pivotal_objective_perturbation, 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.

mle_loss Ran
ran on a synthesized input
gzhu06/generativesourceseparation/generator/glow/commons.py
code served (permissive licence) · get_code("bf8071b60995be0d")
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 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 torch
import math

torch.manual_seed(0)

# Construct tensors for the function
# z: a tensor of latent variables (batch_size=2, dim=4)
z = torch.randn(2, 4)

# logdet: log determinant of Jacobian (batch_size=2, dim=4)
logdet = torch.randn(2, 4)

# mask: binary mask indicating which elements to count (batch_size=2, dim=4)
mask = torch.ones(2, 4)

# m: mean parameter (scalar float)
m = 0.0

ARGS = {
    "z": z,
    "logdet": logdet,
    "mask": mask,
    "m": m,
}
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 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
mle_loss_full Ran
ran on a synthesized input
gzhu06/generativesourceseparation/generator/glow/commons.py
code served (permissive licence) · get_code("6ddfe21dd2c01cb4")
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 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 torch
import math

torch.manual_seed(0)

# Analyze the function to understand what's needed:
# - z: tensor used in arithmetic operations (z - m) ** 2
# - m: tensor, mean parameter
# - logs: tensor, log scale parameter
# - logdet: tensor, log determinant
# - mask: tensor, mask for normalization

# Create small tensors with distinct shapes to allow probing
batch_size = 2
dim = 4

z = torch.randn(batch_size, dim)
m = torch.randn(batch_size, dim)
logs = torch.randn(batch_size, dim)
logdet = torch.randn(batch_size, dim)
mask = torch.ones(batch_size, dim)

ARGS = {
    "z": z,
    "m": m,
    "logs": logs,
    "logdet": logdet,
    "mask": mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 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
resynthesize_from_spec Ran
ran on a synthesized input
gzhu06/generativesourceseparation/inverse_utils.py
code served (permissive licence) · get_code("3ba3d28bff4539ee")
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 (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 torch
import numpy as np
from torch import nn

torch.manual_seed(0)

class MockSTFT(nn.Module):
    def __init__(self):
        super().__init__()
    
    def transform(self, x):
        # x shape: (batch, samples)
        # Returns (magnitude, phase)
        batch_size = x.shape[0]
        magnitude = torch.randn(batch_size, 4, 3)
        phase = torch.randn(batch_size, 4, 3)
        return magnitude, phase
    
    def inverse(self, mag, phase):
        # mag shape: (batch, freq_bins, time_frames)
        # phase shape: (batch, freq_bins, time_frames)
        # Returns reconstructed waveform (batch, samples)
        batch_size = mag.shape[0]
        return torch.randn(batch_size, 512)

class MockSTFTWrapper:
    def __init__(self):
        self.stft_fn = MockSTFT()

# Create test data
specIPT = torch.randn(4, 3)  # (freq_bins, time_frames)
mixWav = np.random.randn(512).astype(np.float32)  # audio samples
stft = MockSTFTWrapper()

ARGS = {
    "specIPT": specIPT,
    "mixWav": mixWav,
    "stft": stft,
}
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 ndarray, ndim 1, dtype family float, shape (512,) — shape (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

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 4 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
intersperse 13 other papers
3 dated earlier, 10 later
2005.11129 (2020-05)
Glow-TTS: A Generative Flow for Text-to-Speech via Monotonic Alignment Search
2105.06337 (2021-05)
Grad-TTS: A Diffusion Probabilistic Model for Text-to-Speech
2106.06103 (2021-06)
Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech
and 10 more not named here

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

Fully-supervised models for source separation are trained on parallel mixture-source data and are currently state-of-the-art. However, such parallel data is often difficult to obtain, and it is cumbersome to adapt trained models to mixtures with new sources. Source-only supervised models, in contrast, only require individual source data for training. In this paper, we first leverage flow-based generators to train individual music source priors and then use these models, along with likelihood-based objectives, to separate music mixtures. We show that in singing voice separation and music separation tasks, our proposed method is competitive with a fully-supervised approach. We also demonstrate that we can flexibly add new types of sources, whereas fully-supervised approaches would require retraining of the entire model.

For agents

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

get_harvested_code_for_paper("2204.09079")
get_code_for_paper("2204.09079")
have("2204.09079")

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

curl https://syntology.ai/api/ran/2204.09079.json

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

[![Syntology run record](https://syntology.ai/api/ran/2204.09079.svg)](https://syntology.ai/paper/2204.09079)

Connect an agent — have() is free.