SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2302.14748 · 2023

Reducing the Prior Mismatch of Stochastic Differential Equations for Diffusion-based Speech Enhancement

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 11 functions out of this paper's own repositories and ran 10 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
sp-uhh/sgmse-bbed canonical 10 of 11
FunctionStatusWhere it lives
energy_ratios Ran sp-uhh/sgmse-bbed/sgmse/util/other.py
code served (permissive licence) · get_code("e4185b3ed3a097b6")
get_act Ran sp-uhh/sgmse-bbed/sgmse/backbones/ncsnpp_utils/layers.py
code served (permissive licence) · get_code("216c782f9d07ef3a")
get_activation Ran sp-uhh/sgmse-bbed/sgmse/backbones/dcunet.py
code served (permissive licence) · get_code("eb5ba43dadcbda84")
get_window Ran sp-uhh/sgmse-bbed/sgmse/data_module.py
code served (permissive licence) · get_code("9e802d63089d5d62")
make_unet_encoder_decoder_args Ran sp-uhh/sgmse-bbed/sgmse/backbones/dcunet.py
code served (permissive licence) · get_code("9ce3b9f139bb58ff")
mean_conf_int Ran sp-uhh/sgmse-bbed/sgmse/util/other.py
code served (permissive licence) · get_code("f36a0fc077e7eeca")
si_sdr_components Ran sp-uhh/sgmse-bbed/sgmse/util/other.py
code served (permissive licence) · get_code("5f57b9e7cfd0a20d")
torch_complex_from_reim Ran sp-uhh/sgmse-bbed/sgmse/backbones/shared.py
code served (permissive licence) · get_code("63cec12b7bc5275c")
unet_decoder_args Ran sp-uhh/sgmse-bbed/sgmse/backbones/dcunet.py
code served (permissive licence) · get_code("0eabe078d2ed0794")
variance_scaling Ran sp-uhh/sgmse-bbed/sgmse/backbones/ncsnpp_utils/layers.py
code served (permissive licence) · get_code("3ac11bfe0ae80195")
ncsn_conv1x1 Not yet run sp-uhh/sgmse-bbed/sgmse/backbones/ncsnpp_utils/layers.py
code served (permissive licence) · get_code("85b2235eb7afaa4f")

Repositories linked to this paper

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

Abstract

Recently, score-based generative models have been successfully employed for the task of speech enhancement. A stochastic differential equation is used to model the iterative forward process, where at each step environmental noise and white Gaussian noise are added to the clean speech signal. While in limit the mean of the forward process ends at the noisy mixture, in practice it stops earlier and thus only at an approximation of the noisy mixture. This results in a discrepancy between the terminating distribution of the forward process and the prior used for solving the reverse process at inference. In this paper, we address this discrepancy and propose a forward process based on a Brownian bridge. We show that such a process leads to a reduction of the mismatch compared to previous diffusion processes. More importantly, we show that our approach improves in objective metrics over the baseline process with only half of the iteration steps and having one hyperparameter less to tune.

For agents

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

get_harvested_code_for_paper("2302.14748")
get_code_for_paper("2302.14748")
have("2302.14748")

Connect an agent — have() is free.