SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1910.09463 · 2019

Using Speech Synthesis to Train End-to-End Spoken Language Understanding Models

arXiv · PDF · Open in the Atlas

Code that ran

9 of 12 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; 1 failure of our instrument; 3 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/1910.09463.json · badge: /api/ran/1910.09463.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
lorenlugosch/end-to-end-SLU canonical (classifier label)
classifier basis: census
3 ran · 1 no recorded run
dscripka/openwakeword community (classifier label)
classifier basis not recorded on the link
5 ran · 2 no recorded run
lorenlugosch/pretrain_speech_model community (classifier label)
classifier basis: plan_row
1 ran · 0 no recorded run
FunctionStatusWhere it lives
flatten_features Ran
ran on a synthesized input
dscripka/openwakeword/openwakeword/custom_verifier_model.py
code served (permissive licence) · get_code("9db28a334416e5df")
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
import torch.nn as nn

torch.manual_seed(0)

# The function takes x, which is iterated over and each element i is flattened
# So x should be a list of tensors
x = [
    torch.randn(2, 4),
    torch.randn(2, 3),
    torch.randn(4, 5)
]

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 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
flip Ran
ran; our contract draft was wrong, not the code
lorenlugosch/pretrain_speech_model/models.py
pointer only (licence: Apache-2.0); 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 (Apache-2.0), 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.

generate_roc_curve_tprs Ran
ran on a synthesized input
dscripka/openwakeword/openwakeword/metrics.py
code served (permissive licence) · get_code("33313e78c0675262")
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 numpy as np
import torch

# Construct valid arguments for generate_roc_curve_tprs
# The function expects:
# - scores: list of scores that supports >= comparison and len()
# - n_points: int for number of threshold points

torch.manual_seed(0)

# Create a list of numpy array scores (simulating prediction scores)
# Using numpy array to support vectorized >= comparison as the code does
scores = np.array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.95])

ARGS = {
    "scores": scores,
    "n_points": 25
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, tqdm — 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
has_params Ran
ran on a synthesized input
lorenlugosch/end-to-end-SLU/models.py
code served (permissive licence) · get_code("29a834b140b9fee7")
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 bool, 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 torch.nn as nn

torch.manual_seed(0)

# Create a simple neural network layer with parameters
class SimpleLayer(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

layer_with_params = SimpleLayer()

ARGS = {
    "layer": layer_with_params
}
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 bool, 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
re_arg Ran
ran on a synthesized input
dscripka/openwakeword/openwakeword/utils.py
code served (permissive licence) · get_code("bccf5e2ef3b3fc2c")
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 function. 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 logging
from functools import wraps

torch.manual_seed(0)

# The function re_arg is a decorator factory that takes a kwarg_map
# and returns a decorator that wraps a function to remap keyword arguments

def sample_function(x, y, z=None):
    """A sample function to be decorated"""
    return f"x={x}, y={y}, z={z}"

# Create a kwarg_map that maps old argument names to new ones
kwarg_map = {
    "old_param_1": "new_param_1",
    "old_param_2": "new_param_2",
    "deprecated_arg": "current_arg"
}

ARGS = {
    "kwarg_map": kwarg_map
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
rms_energy Ran
ran on a synthesized input
lorenlugosch/end-to-end-SLU/data.py
code served (permissive licence) · get_code("fdfd505e22326547")
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 float64, ndim 0, 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 numpy as np

def rms_energy(x):
    return 10 * np.log10((1e-12 + x.dot(x)) / len(x))

# The function takes a numpy array x and:
# - calls x.dot(x) to compute the dot product
# - calls len(x) to get the length
# This means x should be a 1D numpy array

x = np.array([0.1, 0.2, 0.3, 0.4])

ARGS = {
    'x': x
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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 on a synthesized input
lorenlugosch/end-to-end-SLU/models.py
code served (permissive licence) · get_code("be70acdc3215c288")
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 (9,) 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)

# Analyze the sinc function:
# - band: used in arithmetic (2 * math.pi * band * t_right), so a scalar float or tensor
# - t_right: a tensor that supports .is_cuda, torch.sin, division, torch.cat
# The function creates a sinc filter by computing sin(2*pi*band*t)/(...) 

# t_right should be a 1D tensor of time values
band = 0.5  # scalar band parameter
t_right = torch.linspace(0.1, 2.0, 4)  # 1D tensor of 4 time samples, avoid 0 to prevent division issues

ARGS = {
    "band": band,
    "t_right": t_right,
}
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 (9,) — shape (9,) 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
stack_clips Ran
ran on a synthesized input
dscripka/openwakeword/openwakeword/data.py
code served (permissive licence) · get_code("2d07a7c28d9ef33e")
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 (2, 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

# Create sample audio data - list of 1D numpy arrays with different lengths
audio_clip_1 = np.random.randn(8000).astype(np.float32)
audio_clip_2 = np.random.randn(12000).astype(np.float32)
audio_clip_3 = np.random.randn(5000).astype(np.float32)

audio_data = [audio_clip_1, audio_clip_2, audio_clip_3]

ARGS = {
    "audio_data": audio_data,
    "clip_size": 16000
}
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 (2, 16000) — shape (2, 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
train_verifier_model Ran
ran on a synthesized input
dscripka/openwakeword/openwakeword/custom_verifier_model.py
code served (permissive licence) · get_code("09b738a34151425e")
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 Pipeline. 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

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

# Create sample features: N x M array where N=number of examples, M=number of features
# Using shape (2, 4) - 2 examples, 4 features each
# The function expects features to be flattened by flatten_features, which handles lists
# So we create a list of 2D arrays that will be flattened
features = np.array([
    np.random.randn(2, 2),  # First example: 2x2 array
    np.random.randn(2, 2)   # Second example: 2x2 array
], dtype=object)

# Create sample labels: 1D array with binary labels (0 or 1)
# Should have same length as number of examples
labels = np.array([0, 1])

ARGS = {
    "features": features,
    "labels": labels
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Pipeline — a Pipeline output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Pipeline 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
generate_roc_curve_fprs Not yet run
unverified (harvested, no recorded run)
dscripka/openwakeword/openwakeword/metrics.py
code served (permissive licence) · get_code("dabffaf05575f6d7")
How it ran

no invocation recorded.

get_false_positives Not yet run
unverified (harvested, no recorded run)
dscripka/openwakeword/openwakeword/metrics.py
code served (permissive licence) · get_code("65bb5581084bcf53")
How it ran

no invocation recorded.

read_config Not yet run
unverified (harvested, no recorded run)
lorenlugosch/end-to-end-SLU/data.py
code served (permissive licence) · get_code("3a8129d8532e445f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 1 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.

FunctionAlso present inEarliest-datedOthers named
flip 4 other papers
3 dated earlier, 1 later
1703.09452 (2017-03)
SEGAN: Speech Enhancement Generative Adversarial Network
1808.00158 (2018-08)
Speaker Recognition from Raw Waveform with SincNet
1810.12735 (2018-10)
Spoken Language Understanding on the Edge
and 1 more not named here
has_params 1 other paper
1 dated earlier
1810.12735 (2018-10)
Spoken Language Understanding on the Edge
rms_energy 1 other paper
1 dated earlier
1810.12735 (2018-10)
Spoken Language Understanding on the Edge
sinc 1 other paper
1 dated earlier
1810.12735 (2018-10)
Spoken Language Understanding on the Edge

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

End-to-end models are an attractive new approach to spoken language understanding (SLU) in which the meaning of an utterance is inferred directly from the raw audio without employing the standard pipeline composed of a separately trained speech recognizer and natural language understanding module. The downside of end-to-end SLU is that in-domain speech data must be recorded to train the model. In this paper, we propose a strategy for overcoming this requirement in which speech synthesis is used to generate a large synthetic training dataset from several artificial speakers. Experiments on two open-source SLU datasets confirm the effectiveness of our approach, both as a sole source of training data and as a form of data augmentation.

For agents

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

get_harvested_code_for_paper("1910.09463")
get_code_for_paper("1910.09463")
have("1910.09463")

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

curl https://syntology.ai/api/ran/1910.09463.json

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

[![Syntology run record](https://syntology.ai/api/ran/1910.09463.svg)](https://syntology.ai/paper/1910.09463)

Connect an agent — have() is free.