SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2005.05106 · 2020

Multi-band MelGAN: Faster Waveform Generation for High-Quality Text-to-Speech

arXiv · PDF · Open in the Atlas

Code that ran

2 of 7 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 0 have no contract verdict; 2 failures of our instrument; 5 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/2005.05106.json · badge: /api/ran/2005.05106.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
rishikksh20/melgan community (classifier label)
classifier basis not recorded on the link
2 ran · 1 no recorded run
yanggeng1995/FB-MelGAN reimplementation (classifier label)
classifier basis: classifier_2025
0 ran · 1 no recorded run
copy not recorded community
not in the paper's declared repositories
0 ran · 1 no recorded run
Moon-sung-woo/ParallelWaveGan_korean community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
rishikksh20/VocGAN community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
FunctionStatusWhere it lives
dynamic_range_compression Ran
ran; our contract draft was wrong, not the code
rishikksh20/melgan/utils/audio_processing.py
code served (permissive licence) · get_code("83e2c862f7374ac7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 4) 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

torch.manual_seed(0)

# Create a small tensor with positive values for dynamic range compression
# This simulates magnitude spectrogram or similar audio processing data
x = torch.rand(2, 4) * 10  # Random values between 0 and 10

# C is compression factor (default 1)
C = 1

# clip_val is minimum clipping value (default 1e-05)
clip_val = 1e-05

ARGS = {
    'x': x,
    'C': C,
    'clip_val': clip_val
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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 2, dtype family float, shape (2, 4) — shape (2, 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
griffin_lim Ran
ran; our contract draft was wrong, not the code
rishikksh20/melgan/utils/audio_processing.py
code served (permissive licence) · get_code("9f9ec26d4cdfdf7d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 11) 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

torch.manual_seed(0)

class STFT:
    def __init__(self):
        pass
    
    def transform(self, signal):
        # Returns (magnitudes, angles)
        # signal shape: (batch, time)
        batch_size = signal.size(0)
        time_len = signal.size(1)
        # Return magnitudes and angles with shape (batch, freq, time_frames)
        freq_bins = 5
        time_frames = 7
        magnitudes = torch.abs(torch.randn(batch_size, freq_bins, time_frames))
        angles = torch.randn(batch_size, freq_bins, time_frames)
        return magnitudes, angles
    
    def inverse(self, magnitudes, angles):
        # magnitudes shape: (batch, freq, time_frames)
        # angles shape: (batch, freq, time_frames)
        # Returns signal with shape (batch, 1, time)
        batch_size = magnitudes.size(0)
        time_len = 11
        signal = torch.randn(batch_size, 1, time_len)
        return signal

stft_fn = STFT()

# Create magnitudes tensor with shape (batch=2, freq=5, time_frames=7)
magnitudes = torch.abs(torch.randn(2, 5, 7))

ARGS = {
    'magnitudes': magnitudes,
    'stft_fn': stft_fn,
    'n_iters': 3
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 11) — shape (2, 11) 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
attempt_to_restore Not yet run
unverified (harvested, no recorded run)
yanggeng1995/FB-MelGAN/pre_train.py
pointer only (licence: NONE) · get_code("9368459b15367fe4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, yanggeng1995/FB-MelGAN 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. No invocation recorded.

load_checkpoint Not yet run
unverified (harvested, no recorded run)
this paper's copy was not recorded; identical code first harvested from yanggeng1995/GAN-TTS
pointer only; the stored copy get_code serves, from yanggeng1995/GAN-TTS, is pointer only (licence: NONE) · get_code("f443ef0155d2d898")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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. No invocation recorded.

stft Not yet run
unverified (harvested, no recorded run)
rishikksh20/melgan/utils/stft_loss.py
code served (permissive licence) · get_code("d40b7ea72ed329ed")
How it ran

no invocation recorded.

stft Not yet run
unverified (harvested, no recorded run)
Moon-sung-woo/ParallelWaveGan_korean/parallel_wavegan/losses/stft_loss.py
code served (permissive licence) · get_code("d014b1a7520bc12b")
How it ran

no invocation recorded.

stft Not yet run
unverified (harvested, no recorded run)
rishikksh20/VocGAN/utils/stft_loss.py
code served (permissive licence) · get_code("bebea9f2516460be")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 2 the same month, 3 later) and 2 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
griffin_lim 14 other papers
7 dated earlier, 2 the same month, 5 later
1803.09017 (2018-03)
Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis
1811.02122 (2018-11)
Robust and fine-grained prosody control of end-to-end speech synthesis
1812.04342 (2018-12)
Learning latent representations for style control and transfer in end-to-end speech synthesis
and 11 more not named here
dynamic_range_compression 13 other papers
8 dated earlier, 2 the same month, 3 later
1803.09017 (2018-03)
Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis
1811.02122 (2018-11)
Robust and fine-grained prosody control of end-to-end speech synthesis
1812.04342 (2018-12)
Learning latent representations for style control and transfer in end-to-end speech synthesis
and 10 more not named here
load_checkpoint 3 other papers
3 dated earlier
1909.11646 (2019-09)
High Fidelity Speech Synthesis with Adversarial Networks
1910.06711 (2019-10)
MelGAN: Generative Adversarial Networks for Conditional Waveform Synthesis
1910.11480 (2019-10)
Parallel WaveGAN: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogram
stft 3 other papers
1 dated earlier, 2 later
1910.11480 (2019-10)
Parallel WaveGAN: A fast waveform generation model based on generative adversarial networks with multi-resolution spectrogram
2010.15508 (2020-10)
FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement
2106.07889 (2021-06)
UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation

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 propose multi-band MelGAN, a much faster waveform generation model targeting to high-quality text-to-speech. Specifically, we improve the original MelGAN by the following aspects. First, we increase the receptive field of the generator, which is proven to be beneficial to speech generation. Second, we substitute the feature matching loss with the multi-resolution STFT loss to better measure the difference between fake and real speech. Together with pre-training, this improvement leads to both better quality and better training stability. More importantly, we extend MelGAN with multi-band processing: the generator takes mel-spectrograms as input and produces sub-band signals which are subsequently summed back to full-band signals as discriminator input. The proposed multi-band MelGAN has achieved high MOS of 4.34 and 4.22 in waveform generation and TTS, respectively. With only 1.91M parameters, our model effectively reduces the total computational complexity of the original MelGAN from 5.85 to 0.95 GFLOPS. Our Pytorch implementation, which will be open-resourced shortly, can achieve a real-time factor of 0.03 on CPU without hardware specific optimization.

For agents

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

get_harvested_code_for_paper("2005.05106")
get_code_for_paper("2005.05106")
have("2005.05106")

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

curl https://syntology.ai/api/ran/2005.05106.json

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

[![Syntology run record](https://syntology.ai/api/ran/2005.05106.svg)](https://syntology.ai/paper/2005.05106)

Connect an agent — have() is free.