microsoft
/
DNS-Challenge
зеркало из https://github.com/microsoft/DNS-Challenge.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

intial changes for V5 challenge

This commit is contained in:
Hari Dubey 2022-12-06 19:50:39 +00:00
Родитель 5582dcf5ba
Коммит b5e1e13685
9 изменённых файлов: 407 добавлений и 137 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

0
V5_dev_testset/Track1_Headset/embeddings/README.txt Normal file
Просмотреть файл

0
V5_dev_testset/Track1_Headset/transcripts/README.txt Normal file
Просмотреть файл

0
V5_dev_testset/Track2_Speakerphone/embeddings/README.txt Normal file
Просмотреть файл

0
V5_dev_testset/Track2_Speakerphone/transcripts/README.txt Normal file
Просмотреть файл

0
download-dns-challenge-4-pdns.ps1 → download-dns-challenge-5-headset.ps1
Просмотреть файл

0
download-dns-challenge-4-pdns.sh → download-dns-challenge-5-headset.sh
Просмотреть файл

136
download-dns-challenge-5-speakerphone.ps1 Normal file
Просмотреть файл

@ -0,0 +1,136 @@
#!/usr/bin/bash
# ***** Datasets for ICASSP 2022 DNS Challenge 4 - Personalized DNS Track *****
# NOTE: Before downloading, make sure you have enough space
# on your local storage!
# In all, you will need about 380TB to store the UNPACKED data.
# Archived, the same data takes about 200GB total.
# Please comment out the files you don't need before launching
# the script.
# NOTE: By default, the script *DOES NOT* DOWNLOAD ANY FILES!
# Please scroll down and edit this script to pick the
# downloading method that works best for you.
# -------------------------------------------------------------
# The directory structure of the unpacked data is:
# . 362G
# +-- datasets_fullband 64G
# | +-- impulse_responses 5.9G
# | \-- noise_fullband 58G
# +-- pdns_training_set 294G
# | +-- enrollment_embeddings 115M
# | +-- enrollment_wav 42G
# | +-- raw/clean 252G
# | +-- english 168G
# | +-- french 2.1G
# | +-- german 53G
# | +-- italian 17G
# | +-- russian 6.8G
# | \-- spanish 5.4G
# \-- personalized_dev_testset 3.3G
$BLOB_NAMES=@(
"pdns_training_set/raw/pdns_training_set.raw.clean.english_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_001.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_002.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_003.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_004.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_005.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_006.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_007.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_008.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_009.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_010.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_011.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_012.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_013.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_014.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_015.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_016.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_017.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_018.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_019.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.english_020.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.french_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_001.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_002.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_003.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_004.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_005.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_006.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_007.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.german_008.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.italian_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.italian_001.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.italian_002.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.russian_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.spanish_000.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.spanish_001.tar.bz2",
"pdns_training_set/raw/pdns_training_set.raw.clean.spanish_002.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_000.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_001.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_002.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_003.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_004.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.french_000.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.german_000.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.german_001.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.italian_000.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.russian_000.tar.bz2",
"pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.spanish_000.tar.bz2",
"pdns_training_set/pdns_training_set.enrollment_embeddings_000.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_000.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_001.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_002.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_003.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_004.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_005.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_006.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_000.tar.bz2",
"datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_001.tar.bz2",
"datasets_fullband/datasets_fullband.impulse_responses_000.tar.bz2",
"personalized_dev_testset/personalized_dev_testset.enrollment.tar.bz2",
"personalized_dev_testset/personalized_dev_testset.noisy_testclips.tar.bz2"
)
###############################################################
$AZURE_URL="https://dns4public.blob.core.windows.net/dns4archive"
$OUTPUT_PATH="."
mkdir -Force $OUTPUT_PATH/pdns_training_set/raw 2> $null
mkdir -Force $OUTPUT_PATH/pdns_training_set/enrollment_wav 2> $null
mkdir -Force $OUTPUT_PATH/datasets_fullband/noise_fullband 2> $null
foreach ($BLOB in $BLOB_NAMES) {
$URL="$AZURE_URL/$BLOB"
Write-Output "Download: $BLOB"
# DRY RUN: print HTTP response and Content-Length
# WITHOUT downloading the files
curl -s -I "$URL" | head -n 2
# Actually download the files: UNCOMMENT when ready to download
# curl "$URL" -o "$OUTPUT_PATH/$BLOB"
# Same as above, but using wget
# wget "$URL" -O "$OUTPUT_PATH/$BLOB"
# Same, + unpack files on the fly
# curl "$URL" | tar -C "$OUTPUT_PATH" -f - -x -j
}

134
download-dns-challenge-5-speakerphone.sh Normal file
Просмотреть файл

@ -0,0 +1,134 @@
#!/usr/bin/bash
# ***** Datasets for ICASSP 2022 DNS Challenge 4 - Personalized DNS Track *****
# NOTE: Before downloading, make sure you have enough space
# on your local storage!
# In all, you will need about 380TB to store the UNPACKED data.
# Archived, the same data takes about 200GB total.
# Please comment out the files you don't need before launching
# the script.
# NOTE: By default, the script *DOES NOT* DOWNLOAD ANY FILES!
# Please scroll down and edit this script to pick the
# downloading method that works best for you.
# -------------------------------------------------------------
# The directory structure of the unpacked data is:
# . 362G
# +-- datasets_fullband 64G
# | +-- impulse_responses 5.9G
# | \-- noise_fullband 58G
# +-- pdns_training_set 294G
# | +-- enrollment_embeddings 115M
# | +-- enrollment_wav 42G
# | +-- raw/clean 252G
# | +-- english 168G
# | +-- french 2.1G
# | +-- german 53G
# | +-- italian 17G
# | +-- russian 6.8G
# | \-- spanish 5.4G
# \-- personalized_dev_testset 3.3G
BLOB_NAMES=(
pdns_training_set/raw/pdns_training_set.raw.clean.english_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_001.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_002.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_003.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_004.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_005.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_006.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_007.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_008.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_009.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_010.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_011.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_012.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_013.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_014.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_015.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_016.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_017.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_018.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_019.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.english_020.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.french_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_001.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_002.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_003.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_004.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_005.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_006.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_007.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.german_008.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.italian_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.italian_001.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.italian_002.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.russian_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.spanish_000.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.spanish_001.tar.bz2
pdns_training_set/raw/pdns_training_set.raw.clean.spanish_002.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_000.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_001.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_002.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_003.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.english_004.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.french_000.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.german_000.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.german_001.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.italian_000.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.russian_000.tar.bz2
pdns_training_set/enrollment_wav/pdns_training_set.enrollment_wav.spanish_000.tar.bz2
pdns_training_set/pdns_training_set.enrollment_embeddings_000.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_000.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_001.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_002.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_003.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_004.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_005.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_006.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_000.tar.bz2
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_001.tar.bz2
datasets_fullband/datasets_fullband.impulse_responses_000.tar.bz2
personalized_dev_testset/personalized_dev_testset.enrollment.tar.bz2
personalized_dev_testset/personalized_dev_testset.noisy_testclips.tar.bz2
)
###############################################################
AZURE_URL="https://dns4public.blob.core.windows.net/dns4archive"
OUTPUT_PATH="."
mkdir -p $OUTPUT_PATH/{pdns_training_set/{raw,enrollment_wav},datasets_fullband/noise_fullband}
for BLOB in ${BLOB_NAMES[@]}
do
URL="$AZURE_URL/$BLOB"
echo "Download: $BLOB"
# DRY RUN: print HTTP response and Content-Length
# WITHOUT downloading the files
curl -s -I "$URL" | head -n 2
# Actually download the files: UNCOMMENT when ready to download
# curl "$URL" -o "$OUTPUT_PATH/$BLOB"
# Same as above, but using wget
# wget "$URL" -O "$OUTPUT_PATH/$BLOB"
# Same, + unpack files on the fly
# curl "$URL" | tar -C "$OUTPUT_PATH" -f - -x -j
done

274
pdns_synthesizer_icassp2022.cfg → pdns_synthesizer_icassp2023.cfg
Просмотреть файл

@ -1,137 +1,137 @@
# Configuration for generating Noisy Speech Dataset # Configuration for generating Noisy Speech Dataset
# - sampling_rate: Specify the sampling rate. Default is 16 kHz # - sampling_rate: Specify the sampling rate. Default is 16 kHz
# - audioformat: default is .wav # - audioformat: default is .wav
# - audio_length: Minimum Length of each audio clip (noisy and clean speech) in seconds that will be generated by augmenting utterances. # - audio_length: Minimum Length of each audio clip (noisy and clean speech) in seconds that will be generated by augmenting utterances.
# - silence_length: Duration of silence introduced between clean speech utterances. # - silence_length: Duration of silence introduced between clean speech utterances.
# - total_hours: Total number of hours of data required. Units are in hours. # - total_hours: Total number of hours of data required. Units are in hours.
# - snr_lower: Lower bound for SNR required (default: 0 dB) # - snr_lower: Lower bound for SNR required (default: 0 dB)
# - snr_upper: Upper bound for SNR required (default: 40 dB) # - snr_upper: Upper bound for SNR required (default: 40 dB)
# - target_level_lower: Lower bound for the target audio level before audiowrite (default: -35 dB) # - target_level_lower: Lower bound for the target audio level before audiowrite (default: -35 dB)
# - target_level_upper: Upper bound for the target audio level before audiowrite (default: -15 dB) # - target_level_upper: Upper bound for the target audio level before audiowrite (default: -15 dB)
# - total_snrlevels: Number of SNR levels required (default: 5, which means there are 5 levels between snr_lower and snr_upper) # - total_snrlevels: Number of SNR levels required (default: 5, which means there are 5 levels between snr_lower and snr_upper)
# - clean_activity_threshold: Activity threshold for clean speech # - clean_activity_threshold: Activity threshold for clean speech
# - noise_activity_threshold: Activity threshold for noise # - noise_activity_threshold: Activity threshold for noise
# - fileindex_start: Starting file ID that will be used in filenames # - fileindex_start: Starting file ID that will be used in filenames
# - fileindex_end: Last file ID that will be used in filenames # - fileindex_end: Last file ID that will be used in filenames
# - is_test_set: Set it to True if it is the test set, else False for the training set # - is_test_set: Set it to True if it is the test set, else False for the training set
# - noise_dir: Specify the directory path to all noise files # - noise_dir: Specify the directory path to all noise files
# - Speech_dir: Specify the directory path to all clean speech files # - Speech_dir: Specify the directory path to all clean speech files
# - noisy_destination: Specify path to the destination directory to store noisy speech # - noisy_destination: Specify path to the destination directory to store noisy speech
# - clean_destination: Specify path to the destination directory to store clean speech # - clean_destination: Specify path to the destination directory to store clean speech
# - noise_destination: Specify path to the destination directory to store noise speech # - noise_destination: Specify path to the destination directory to store noise speech
# - log_dir: Specify path to the directory to store all the log files # - log_dir: Specify path to the directory to store all the log files
# Configuration for unit tests # Configuration for unit tests
# - snr_test: Set to True if SNR test is required, else False # - snr_test: Set to True if SNR test is required, else False
# - norm_test: Set to True if Normalization test is required, else False # - norm_test: Set to True if Normalization test is required, else False
# - sampling_rate_test: Set to True if Sampling Rate test is required, else False # - sampling_rate_test: Set to True if Sampling Rate test is required, else False
# - clipping_test: Set to True if Clipping test is required, else False # - clipping_test: Set to True if Clipping test is required, else False
# - unit_tests_log_dir: Specify path to the directory where you want to store logs # - unit_tests_log_dir: Specify path to the directory where you want to store logs
[noisy_speech] [noisy_speech]
sampling_rate: 48000 sampling_rate: 48000
audioformat: *.wav audioformat: *.wav
audio_length: 30 audio_length: 30
# 15, 12, 30 # 15, 12, 30
silence_length: 0.2 silence_length: 0.2
total_hours: 1000 total_hours: 1000
# 1000 # 1000
#200 #200
# 2.5, 500, 100 # 2.5, 500, 100
snr_lower: -5 snr_lower: -5
#-5, 0 #-5, 0
snr_upper: 20 snr_upper: 20
# 25, 40 # 25, 40
randomize_snr: True randomize_snr: True
target_level_lower: -35 target_level_lower: -35
target_level_upper: -15 target_level_upper: -15
total_snrlevels: 31 total_snrlevels: 31
# 5 # 5
clean_activity_threshold: 0.0 clean_activity_threshold: 0.0
noise_activity_threshold: 0.2 noise_activity_threshold: 0.2
fileindex_start: None fileindex_start: None
fileindex_end: None fileindex_end: None
is_test_set: False is_test_set: False
# True, False # True, False
noise_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/noise noise_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/noise
#/mnt/f/4th_DNSChallenge/INTERSPEECH_2021/DNS-Challenge/datasets_fullband/noise #/mnt/f/4th_DNSChallenge/INTERSPEECH_2021/DNS-Challenge/datasets_fullband/noise
#F:\4th_DNSChallenge\INTERSPEECH_2021\DNS-Challenge\datasets_fullband\noise #F:\4th_DNSChallenge\INTERSPEECH_2021\DNS-Challenge\datasets_fullband\noise
#datasets\pdns_training_set\noise #datasets\pdns_training_set\noise
#\test_set2\Test_Noise #\test_set2\Test_Noise
# datasets\noise # datasets\noise
# \datasets\noise # \datasets\noise
speech_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/clean speech_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/clean
# D:\kanhawin_git\primary_speakers_VCTK_16k_for_synthesizer # D:\kanhawin_git\primary_speakers_VCTK_16k_for_synthesizer
# datasets\test_set2\Singing_Voice\wav_16k # datasets\test_set2\Singing_Voice\wav_16k
# dir with secondary speaker clean speech # dir with secondary speaker clean speech
speech_dir2: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/clean speech_dir2: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/clean
#D:\kanhawin_git\secondary_speakers_voxCeleb2_16k #D:\kanhawin_git\secondary_speakers_voxCeleb2_16k
# datasets\test_set2\Singing_Voice\wav_16k # datasets\test_set2\Singing_Voice\wav_16k
spkid_csv: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/filelists/complete_ps_split.csv spkid_csv: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/filelists/complete_ps_split.csv
#/mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/filelists/vctk_spkid.csv #/mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/filelists/vctk_spkid.csv
# datasets\clean # datasets\clean
noise_types_excluded: None noise_types_excluded: None
rir_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/impulse_responses rir_dir: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/impulse_responses
#/mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/impulse_responses #/mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/impulse_responses
# F:\4th_DNSChallenge\ICASSP_2022\DNS-Challenge\datasets\impulse_responses # F:\4th_DNSChallenge\ICASSP_2022\DNS-Challenge\datasets\impulse_responses
# \datasets\clean # \datasets\clean
noisy_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/noisy noisy_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/noisy
# datasets/training_data/noisy # datasets/training_data/noisy
# datasets\test_set2\synthetic_personalizeddns\noisy # datasets\test_set2\synthetic_personalizeddns\noisy
#training_set2_onlyrealrir\noisy #training_set2_onlyrealrir\noisy
#\noisy #\noisy
clean_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/clean clean_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/clean
#datasets\test_set2\synthetic_personalizeddns\clean #datasets\test_set2\synthetic_personalizeddns\clean
# training_set2_onlyrealrir\clean # training_set2_onlyrealrir\clean
# \clean # \clean
noise_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/noise noise_destination: /mnt/f/4th_DNSChallenge/ICASSP_2022/DNS-Challenge/datasets/pdns_training_set/mixed/noise
# datasets/training_data/noise # datasets/training_data/noise
#datasets\test_set2\synthetic_personalizeddns\noise #datasets\test_set2\synthetic_personalizeddns\noise
#training_set2_onlyrealrir\noise #training_set2_onlyrealrir\noise
# \noise # \noise
log_dir: logs log_dir: logs
# \logs # \logs
# Config: add singing voice to clean speech # Config: add singing voice to clean speech
clean_singing: datasets\clean_singing\VocalSet11\FULL clean_singing: datasets\clean_singing\VocalSet11\FULL
singing_choice: 3 singing_choice: 3
# 1 for only male, 2 for only female, 3 (default) for both male and female # 1 for only male, 2 for only female, 3 (default) for both male and female
# Config: add reverb to clean speech # Config: add reverb to clean speech
rir_choice: 1 rir_choice: 1
# 1 for only real rir, 2 for only synthetic rir, 3 (default) use both real and synthetic # 1 for only real rir, 2 for only synthetic rir, 3 (default) use both real and synthetic
lower_t60: 0.3 lower_t60: 0.3
# lower bound of t60 range in seconds # lower bound of t60 range in seconds
upper_t60: 1.3 upper_t60: 1.3
# upper bound of t60 range in seconds # upper bound of t60 range in seconds
rir_table_csv: datasets\acoustic_params\RIR_table_simple.csv rir_table_csv: datasets\acoustic_params\RIR_table_simple.csv
clean_speech_t60_csv: datasets\acoustic_params\cleanspeech_table_t60_c50.csv clean_speech_t60_csv: datasets\acoustic_params\cleanspeech_table_t60_c50.csv
# percent_for_adding_reverb=0.5 # percentage of clean speech convolved with RIR # percent_for_adding_reverb=0.5 # percentage of clean speech convolved with RIR
# pdns testsets # pdns testsets
# primary_data: D:\kanhawin_git\primary_speakers_VCTK_16k # primary_data: D:\kanhawin_git\primary_speakers_VCTK_16k
#'D:\PersonalizedDNS_dataset\synthetic_primary' #'D:\PersonalizedDNS_dataset\synthetic_primary'
# secondary_data='D:\kanhawin_git\secondary_speakers_voxCeleb2_16k' # secondary_data='D:\kanhawin_git\secondary_speakers_voxCeleb2_16k'
#'D:\PersonalizedDNS_dataset\synthetic_secondary' #'D:\PersonalizedDNS_dataset\synthetic_secondary'
# noise_data= datasets\test_set2\synthetic\noise # noise_data= datasets\test_set2\synthetic\noise
# pdns_testset_clean= datasets\test_set2\pdns\clean # pdns_testset_clean= datasets\test_set2\pdns\clean
# pdns_testset_noisy= datasets\test_set2\pdns\noisy # pdns_testset_noisy= datasets\test_set2\pdns\noisy
# adaptation_data_seconds=120 # adaptation_data_seconds=120
# num_primary_spk=100 # num_primary_spk=100
# num_clips=600 # num_clips=600
# Unit tests config # Unit tests config
snr_test: True snr_test: True
norm_test: True norm_test: True
sampling_rate_test = True sampling_rate_test = True
clipping_test = True clipping_test = True
unit_tests_log_dir: unittests_logs unit_tests_log_dir: unittests_logs