#-------------------------------------------
# Basic information
#-------------------------------------------
team: your_teamname
team_email: email_for_receiving_notification
team_members:
# Member 1
1:
name: member1_name
affiliation: member1_affiliation
# Member 2
2:
name: member2_name
affiliation: member2_affiliation
# Member 3
3:
name: member3_name
affiliation: member3_affiliation
#-------------------------------------------
# Data description
#-------------------------------------------
# e.g., "simulated on the fly using the official configuration"
training_data:
use_additional_rirs: # true or false
additional_rir_info:
# example:
# is_real: True
# source: https://url.to.real.rirs
# number_of_samples: 1000
# RIR 1 (for real RIRs, only those included in the official data list are allowed)
1:
is_real:
source:
number_of_samples:
# RIR 2
2:
is_real:
source:
number_of_samples:
# e.g., "same as official validation data"
validation_data:
#-------------------------------------------
# System description
#-------------------------------------------
# "discriminative", "generative", "discriminative followed by generative", etc.
system_type:
# please briefly describe the architecture here
# e.g., CNN, Transformer.
model_architecture:
# model parameter size in million, e.g., 6.23 [M]
model_size:
# time, time-frequency, etc.
domain:
# please briefly describe loss function here
# e.g., "time-domain si-snr", "time-domain l1 + TF-domain gan", etc.
loss_funcition:
# please briefly describe data augmentation here
# mp3 compression, wind noise, etc.
data_augmentation:
use_pretrained_models: # true or false
pretrained_models:
# example:
# name: HuBERT-Large
# link: https://huggingface.co/facebook/hubert-large-ll60k
# usage: used for converting the input speech signal into discrete tokens
# Pre-trained model 1
1:
name:
link:
usage:
# Pre-trained model 2
2:
name:
link:
usage: