SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2108.02271 · 2021

Daft-Exprt: Cross-Speaker Prosody Transfer on Any Text for Expressive Speech Synthesis

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/2108.02271.json · badge: /api/ran/2108.02271.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
ubisoft/ubisoft-laforge-daft-exprt canonical (classifier label)
classifier basis: census
9 ran · 3 no recorded run
FunctionStatusWhere it lives
check_features_config_used Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/extract_features.py
code served (permissive licence) · get_code("1ed72dcd693adf5f")
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 json
import os
import tempfile
import types

# Create a temporary directory with a config file
temp_dir = tempfile.mkdtemp()
config_file_path = os.path.join(temp_dir, 'config.json')

# Create a sample config that matches FEATURES_HPARAMS
sample_config = {
    'centered': True,
    'cutoff': 0.5,
    'f0_interval': 10,
    'filter_length': 1024,
    'hop_length': 256,
    'language': 'en',
    'mel_fmax': 7600,
    'mel_fmin': 55,
    'min_clipping': 0.1,
    'max_f0': 800,
    'min_f0': 50,
    'n_mel_channels': 80,
    'order': 5,
    'sampling_rate': 22050,
    'symbols': 'abcdefg',
    'uv_cost': 0.5,
    'uv_interval': 20
}

with open(config_file_path, 'w') as f:
    json.dump(sample_config, f)

# Create a SimpleNamespace object with the same parameters for hparams
hparams = types.SimpleNamespace(**sample_config)

ARGS = {
    'features_dir': temp_dir,
    'hparams': hparams
}
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 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
collate_tensors Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/generate.py
code served (permissive licence) · get_code("c671cdd78df04529")
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
import tempfile
import os

torch.manual_seed(0)

# Create a temporary .npz file for references
temp_dir = tempfile.mkdtemp()
ref_file = os.path.join(temp_dir, "ref.npz")

# Create sample reference data
energy_ref = np.random.randn(5).astype(np.float32)
pitch_ref = np.random.randn(5).astype(np.float32)
mel_spec_ref = np.random.randn(80, 5).astype(np.float32)  # n_mel_channels=80, time_steps=5

np.savez(ref_file, energy=energy_ref, pitch=pitch_ref, mel_spec=mel_spec_ref)

# Create hparams config
class HParams:
    def __init__(self):
        self.symbols = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h']
        self.n_mel_channels = 80

hparams = HParams()

# Create batch data
batch_size = 2

# Batch sentences - list of lists containing symbols/phones
batch_sentences = [
    [['a', 'b'], 'c', ['d', 'e']],
    ['a', 'b', 'c']
]

# Duration factors
batch_dur_factors = [
    [1.0, 1.5, 2.0, 1.0, 1.2],
    None
]

# Energy factors
batch_energy_factors = [
    [0.9, 1.0, 1.1, 0.8, 1.0],
    None
]

# Pitch factors
batch_pitch_factors = [
    [0.0, 0.5, 1.0, -0.5, 0.0],
    None
]

# Pitch transform type
pitch_transform = 'add'

# Reference files
batch_refs = [ref_file, ref_file]

# Speaker IDs
batch_speaker_ids = [0, 1]

# File names
batch_file_names = ['file_0.wav', 'file_1.wav']

ARGS = {
    'batch_sentences': batch_sentences,
    'batch_dur_factors': batch_dur_factors,
    'batch_energy_factors': batch_energy_factors,
    'batch_pitch_factors': batch_pitch_factors,
    'pitch_transform': pitch_transform,
    'batch_refs': batch_refs,
    'batch_speaker_ids': batch_speaker_ids,
    'batch_file_names': batch_file_names,
    'hparams': hparams
}
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
duration_to_integer Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/extract_features.py
code served (permissive licence) · get_code("01590cc54d456ab1")
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 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.

class HParams:
    def __init__(self):
        self.sampling_rate = 22050
        self.filter_length = 1024
        self.hop_length = 256
        self.centered = True

hparams = HParams()

float_durations = [
    (0.0, 0.1),
    (0.1, 0.3),
    (0.3, 0.5),
    (0.5, 0.7),
]

nb_samples = None

ARGS = {
    "float_durations": float_durations,
    "hparams": hparams,
    "nb_samples": nb_samples,
}
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 int — 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
expand_abbreviations Ran
ran; our contract draft was wrong, not the code
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/cleaners.py
code served (permissive licence) · get_code("cea1f83e9db28ed8")
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 re

def expand_abbreviations(text):
    _abbreviations = [(re.compile('\\b%s\\.' % x[0], re.IGNORECASE), x[1]) for x in [('mrs', 'misess'), ('mr', 'mister'), ('dr', 'doctor'), ('st', 'saint'), ('co', 'company'), ('jr', 'junior'), ('maj', 'major'), ('gen', 'general'), ('drs', 'doctors'), ('rev', 'reverend'), ('lt', 'lieutenant'), ('hon', 'honorable'), ('sgt', 'sergeant'), ('capt', 'captain'), ('esq', 'esquire'), ('ltd', 'limited'), ('col', 'colonel'), ('ft', 'fort')]]
    for regex, replacement in _abbreviations:
        text = re.sub(regex, replacement, text)
    return text

ARGS = {
    'text': 'Dr. Smith and Mrs. Johnson went to the fort. Mr. Brown is a colonel.'
}
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_min_phone_duration Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/extract_features.py
code served (permissive licence) · get_code("f2f80f2af9c345b2")
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 float, 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 numpy as np

# The function expects:
# - lines: an iterable of strings that can be stripped and split by tab
# - min_phone_dur: a float (default 1000.0)

# Looking at the function body:
# - lines are iterated over
# - each line is stripped and split by '\t'
# - line[0] and line[1] are converted to float (begin and end times)
# - end - begin is compared to min_phone_dur
# - the minimum duration is tracked and returned

# Create sample .markers file format lines
lines = [
    "0.0\t0.05",      # duration: 0.05 (shortest)
    "0.05\t0.15",     # duration: 0.10
    "0.15\t0.35",     # duration: 0.20
    "0.35\t0.65",     # duration: 0.30
]

ARGS = {
    "lines": lines,
    "min_phone_dur": 1000.0
}
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 float, 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
get_symbols_durations Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/features_stats.py
code served (permissive licence) · get_code("da2c10e8cc4ed93a")
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 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 logging
import logging.handlers
import os
import tempfile
import queue

# Create a temporary markers file with valid content
temp_dir = tempfile.mkdtemp()
markers_file = os.path.join(temp_dir, "markers.txt")

# Write sample marker data with tab-separated values
# Format: begin_time, end_time, _, symbol, _, _
with open(markers_file, 'w', encoding='utf-8') as f:
    f.write("0.0\t0.5\tignored1\ta\tignored2\tignored3\n")
    f.write("0.5\t1.2\tignored1\tb\tignored2\tignored3\n")
    f.write("1.2\t1.8\tignored1\ta\tignored2\tignored3\n")

# Create hparams with symbols attribute
class HParams:
    def __init__(self):
        self.symbols = ['a', 'b', 'c']

hparams = HParams()

# Create a log queue for the QueueHandler
log_queue = queue.Queue()

ARGS = {
    'markers_file': markers_file,
    'hparams': hparams,
    'log_queue': log_queue
}
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 2, 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_voiced_pitch_values Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/features_stats.py
code served (permissive licence) · get_code("d27e300dc16a6ffa")
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 logging.handlers
import os
import tempfile
import uuid

# Create a temporary pitch file with some test data
temp_dir = tempfile.mkdtemp()
pitch_file = os.path.join(temp_dir, "test_pitch.txt")

# Write some pitch values to the file (mix of zeros and non-zero values)
with open(pitch_file, 'w', encoding='utf-8') as f:
    f.write("0.0\n")
    f.write("150.5\n")
    f.write("0.0\n")
    f.write("155.3\n")
    f.write("148.2\n")
    f.write("0.0\n")

# Create a queue for logging
log_queue = __import__('queue').Queue()

ARGS = {
    "pitch_file": pitch_file,
    "log_queue": log_queue,
}
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
lowercase Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/cleaners.py
code served (permissive licence) · get_code("296274b940cb0b32")
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 torch

torch.manual_seed(0)

ARGS = {
    "text": "Hello World"
}
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 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
reconstruct_signal_griffin_lim Ran
ran on a synthesized input
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/griffin_lim.py
code served (permissive licence) · get_code("44f18502201dd01e")
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 numpy as np
import logging

# Create a simple logger
logger = logging.getLogger("test_logger")
logger.setLevel(logging.DEBUG)
handler = logging.StreamHandler()
logger.addHandler(handler)

# Create valid arguments for the function
# magnitude_spectrogram should be shape (n_fft // 2 + 1, T)
# Let's use n_fft = 512, so n_fft // 2 + 1 = 257
# And T (time slices) = 10
n_fft = 512
n_freq_bins = n_fft // 2 + 1
time_slices = 10

magnitude_spectrogram = np.random.rand(n_freq_bins, time_slices).astype(np.float32)

step_size = 256  # typical hop length, should be less than n_fft
iterations = 5  # small number for testing

ARGS = {
    "magnitude_spectrogram": magnitude_spectrogram,
    "step_size": step_size,
    "iterations": iterations,
    "logger": logger
}
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 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_non_zero_energy_values Not yet run
unverified (harvested, no recorded run)
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/features_stats.py
code served (permissive licence) · get_code("9e5c0aca1952744a")
How it ran

no invocation recorded.

nnls Not yet run
unverified (harvested, no recorded run)
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/griffin_lim.py
code served (permissive licence) · get_code("9d4bfd799b0f0664")
How it ran

no invocation recorded.

prepare_data_loaders Not yet run
unverified (harvested, no recorded run)
ubisoft/ubisoft-laforge-daft-exprt/src/daft_exprt/data_loader.py
code served (permissive licence) · get_code("0552e22ebe1e5aaf")
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, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 2 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
expand_abbreviations 8 other papers
5 dated earlier, 3 later
1703.10135 (2017-03)
Tacotron: Towards End-to-End Speech Synthesis
2005.01245 (2020-05)
Can Speaker Augmentation Improve Multi-Speaker End-to-End TTS?
2005.05957 (2020-05)
Flowtron: an Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis
and 5 more not named here
lowercase 6 other papers
5 dated earlier, 1 later
1703.10135 (2017-03)
Tacotron: Towards End-to-End Speech Synthesis
2005.01245 (2020-05)
Can Speaker Augmentation Improve Multi-Speaker End-to-End TTS?
2005.11129 (2020-05)
Glow-TTS: A Generative Flow for Text-to-Speech via Monotonic Alignment Search
and 3 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

This paper presents Daft-Exprt, a multi-speaker acoustic model advancing the state-of-the-art for cross-speaker prosody transfer on any text. This is one of the most challenging, and rarely directly addressed, task in speech synthesis, especially for highly expressive data. Daft-Exprt uses FiLM conditioning layers to strategically inject different prosodic information in all parts of the architecture. The model explicitly encodes traditional low-level prosody features such as pitch, loudness and duration, but also higher level prosodic information that helps generating convincing voices in highly expressive styles. Speaker identity and prosodic information are disentangled through an adversarial training strategy that enables accurate prosody transfer across speakers. Experimental results show that Daft-Exprt significantly outperforms strong baselines on inter-text cross-speaker prosody transfer tasks, while yielding naturalness comparable to state-of-the-art expressive models. Moreover, results indicate that the model discards speaker identity information from the prosody representation, and consistently generate speech with the desired voice. We publicly release our code and provide speech samples from our experiments.

For agents

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

get_harvested_code_for_paper("2108.02271")
get_code_for_paper("2108.02271")
have("2108.02271")

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

curl https://syntology.ai/api/ran/2108.02271.json

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

[![Syntology run record](https://syntology.ai/api/ran/2108.02271.svg)](https://syntology.ai/paper/2108.02271)

Connect an agent — have() is free.