remnove the datasets/ and datasets_fullband/ folders and update the
download script
This commit is contained in:
Sergiy Matusevych
Родитель
8362d4c713
Коммит
5b0e929d09
|
|
@ -1,3 +1,5 @@
|
||||||
|
datasets/
|
||||||
|
datasets_fullband/
|
||||||
training_set/
|
training_set/
|
||||||
training_set2/
|
training_set2/
|
||||||
training_set2_onlyrealrir/
|
training_set2_onlyrealrir/
|
||||||
|
|
@ -12,3 +14,6 @@ __pycache__/
|
||||||
*~
|
*~
|
||||||
/.vs/
|
/.vs/
|
||||||
/.vscode/
|
/.vscode/
|
||||||
|
*.wav
|
||||||
|
*.tar.bz2
|
||||||
|
*.zip
|
||||||
|
|
|
||||||
|
|
@ -1,6 +0,0 @@
|
||||||
/clean/
|
|
||||||
/dev_testset/
|
|
||||||
/impulse_responses/
|
|
||||||
/noise/
|
|
||||||
*.tar.bz2
|
|
||||||
*.zip
|
|
||||||
|
|
@ -1,88 +0,0 @@
|
||||||
|
|
||||||
# Deep Noise Suppression (DNS) Challenge 3 - INTERSPEECH 2021
|
|
||||||
|
|
||||||
**NOTE:** This README describes the **PAST** DNS Challenge!
|
|
||||||
|
|
||||||
The data for it is still available, and is described below. If you are interested in the latest DNS Challenge, please refer to the main [README.md](/README.md) file.
|
|
||||||
|
|
||||||
## Wideband Datasets
|
|
||||||
This directory is the default location where the **wideband** datasets will be downloaded to and
|
|
||||||
stored. After the download, you will see the following directory structure:
|
|
||||||
```
|
|
||||||
datasets 229G
|
|
||||||
├── clean 204G
|
|
||||||
│ ├── emotional_speech 403M
|
|
||||||
│ ├── french_data 21G
|
|
||||||
│ ├── german_speech 66G
|
|
||||||
│ ├── italian_speech 14G
|
|
||||||
│ ├── mandarin_speech 21G
|
|
||||||
│ ├── read_speech 61G
|
|
||||||
│ ├── russian_speech 5.1G
|
|
||||||
│ ├── singing_voice 979M
|
|
||||||
│ └── spanish_speech 17G
|
|
||||||
├── dev_testset 211M
|
|
||||||
├── impulse_responses 4.3G
|
|
||||||
│ ├── SLR26 2.1G
|
|
||||||
│ └── SLR28 2.3G
|
|
||||||
└── noise 20G
|
|
||||||
```
|
|
||||||
|
|
||||||
## Downloading the data
|
|
||||||
|
|
||||||
Datasets will be downloaded when you run the `download-dns-challenge-3.sh` script. Note that the
|
|
||||||
data is no longer part of this git repository and git LFS is not required.
|
|
||||||
|
|
||||||
## Datasets for training
|
|
||||||
The [base paper](/docs/ICASSP_2021_DNS_challenge.pdf) is available in this repo and describes the
|
|
||||||
training and the test data sets in detail.
|
|
||||||
|
|
||||||
### Development stage test set
|
|
||||||
* `dev_testset` directory contains the test set that the participants can use during their development phase.
|
|
||||||
<!--
|
|
||||||
* The <i>track 1</i> directory contains both synthetic and real recordings of the test set.
|
|
||||||
* The <i>track 2</i> directory contains both synthetic and real recordings for the personalized DNS track. The <i>adaptation_data</i> directory contains the utterances from each speaker that can be used adapt the noise suppressor to work better for that particular speaker.
|
|
||||||
-->
|
|
||||||
|
|
||||||
### Clean Speech
|
|
||||||
* The clean speech dataset is derived from the public audio books dataset called Librivox.
|
|
||||||
* Librivox has recordings of volunteers reading over 10,000 public domain audio books in various languages, with majority of which are in English. In total, there are 11,350 speakers.
|
|
||||||
* A section of these recordings is of excellent quality, meaning that the speech was recorded using good quality microphones in a silent and less reverberant environments.
|
|
||||||
* But there are many audio recordings that are of poor speech quality with speech distortion, background noise and reverberation. Hence, it is important to filter the data based on speech quality.
|
|
||||||
* We used the online subjective test framework ITU-T P.808 to sort the book chapters by subjective quality.
|
|
||||||
* The audio chapters in Librivox are of variable length ranging from few seconds to several minutes.
|
|
||||||
* We sampled 10 random clips from each book chapter, each 10 seconds in duration. For each clip we had 3 ratings, and the Mean Opinion Score (MOS) across the all clips was used as the book chapter MOS.
|
|
||||||
* The upper quartile with respect to MOS was chosen as our clean speech dataset, which are top 25% of the clips with MOS as a metric.
|
|
||||||
* The upper quartile comprised of audio chapters with 4.3 ≤ MOS ≤ 5. We removed clips from speakers with less than 15 minutes of speech. The resulting dataset has 500 hours of speech from 2150 speakers.
|
|
||||||
* All the filtered clips are then split into segments of 31 seconds.
|
|
||||||
* Singing voice is from VocalSet corpus. It has 10.1 hrs of singing from 20 professional singers.
|
|
||||||
* Emotion speech from diverse ethnic background is provided. Emotions such as Anger, Disgust, Fear, Happy, Neutral, and Sad at four intensity levels: Low, Medium, High, Unspecified are used.
|
|
||||||
* Non-English clips consisting of tonal and non-tonal languages are included.
|
|
||||||
* More details about the clean speech data can be found in our [ICASSP 2021 DNS Challenge paper](/docs/ICASSP_2021_DNS_challenge.pdf).
|
|
||||||
<!--
|
|
||||||
FIXME: The original URL was this:
|
|
||||||
https://github.com/microsoft/DNS-Challenge/blob/icassp21/addrir/docs/ICASSP_2021_deep_noise_suppression_challenge.pdf
|
|
||||||
The branch and the file do not exist in git history. Is that the right URL?
|
|
||||||
-->
|
|
||||||
|
|
||||||
### Noise
|
|
||||||
* The noise clips were selected from Audioset and Freesound.
|
|
||||||
* Audioset is a collection of about 2 million human-labeled 10s sound clips drawn from YouTube videos and belong to about 600 audio events.
|
|
||||||
* Like the Librivox data, certain audio event classes are overrepresented. For example, there are over a million clips with audio classes music and speech and less than 200 clips for classes such as toothbrush, creak etc.
|
|
||||||
* Approximately, 42% of the clips have single class, but the rest may have 2 to 15 labels.
|
|
||||||
* Hence, we developed a sampling approach to balance the dataset in such a way that each class has at least 500 clips.
|
|
||||||
* We also used a speech activity detector (trained classifier) to remove the clips with any kind of speech activity. The reason is to avoid suppression of speech by the noise suppression model trained to suppress speech like noise.
|
|
||||||
* The resulting dataset has about 150 audio classes and 60,000 clips. We also augmented an additional 10,000 noise clips downloaded from Freesound and DEMAND databases.
|
|
||||||
* The chosen noise types are more relevant to VOIP applications.
|
|
||||||
|
|
||||||
### Room Impulse Responses (RIR)
|
|
||||||
* 3076 real and about 115000 synthetic RIRs are provided.
|
|
||||||
* These room impulse responses can be convolved with clean speech to produce reverberant speech.
|
|
||||||
* Participants can simultaneously train for dereverb and denoising.
|
|
||||||
|
|
||||||
### Acoustic Parameters
|
|
||||||
* We provide two acoustic parameters:
|
|
||||||
* (i) Reveberation time, T60, and
|
|
||||||
* (ii) Clarity, C50
|
|
||||||
for all audio clips in clean speech of the training set.
|
|
||||||
* These parameters are supposed to provide flexibility to researchers for choose a subset of
|
|
||||||
provided data for controlled studies.
|
|
||||||
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
|
|
@ -1,5 +0,0 @@
|
||||||
/clean_fullband/
|
|
||||||
/dev_testset_fullband/
|
|
||||||
/noise_fullband/
|
|
||||||
*.tar.bz2
|
|
||||||
*.zip
|
|
||||||
|
|
@ -1,79 +0,0 @@
|
||||||
|
|
||||||
# Deep Noise Suppression (DNS) Challenge 3 - INTERSPEECH 2021
|
|
||||||
|
|
||||||
**NOTE:** This README describes the **PAST** DNS Challenge!
|
|
||||||
|
|
||||||
The data for it is still available, and is described below. If you are interested in the latest DNS Challenge, please refer to the main [README.md](/README.md) file.
|
|
||||||
|
|
||||||
# Fullband datasets
|
|
||||||
This directory is the default location where the **fullband** datasets will be downloaded to and
|
|
||||||
stored. After the download, you will see the following directory structure:
|
|
||||||
```
|
|
||||||
datasets_fullband 600G
|
|
||||||
├── clean_fullband 542G
|
|
||||||
│ ├── VocalSet_48kHz_mono 974M
|
|
||||||
│ ├── emotional_speech 1.2G
|
|
||||||
│ ├── french_data 62G
|
|
||||||
│ ├── german_speech 194G
|
|
||||||
│ ├── italian_speech 42G
|
|
||||||
│ ├── read_speech 182G
|
|
||||||
│ ├── russian_speech 12G
|
|
||||||
│ └── spanish_speech 50G
|
|
||||||
├── dev_testset_fullband 630M
|
|
||||||
└── noise_fullband 58G
|
|
||||||
```
|
|
||||||
|
|
||||||
## Downloading the data
|
|
||||||
Datasets will be downloaded when you run the `download-dns-challenge-3.sh` script. Note that the
|
|
||||||
data is no longer part of this git repository and git LFS is not required.
|
|
||||||
|
|
||||||
## Datasets for training
|
|
||||||
The [base paper](/docs/ICASSP_2021_DNS_challenge.pdf) is available in this repo and describes the
|
|
||||||
training and the test data sets in detail.
|
|
||||||
|
|
||||||
### Development stage test set
|
|
||||||
* `dev_testset_fullband` directory contains the test set that the participants can use during their
|
|
||||||
development phase.
|
|
||||||
<!--
|
|
||||||
* The <i>track 1</i> directory contains both synthetic and real recordings of the test set.
|
|
||||||
* The <i>track 2</i> directory contains both synthetic and real recordings for the personalized DNS track. The <i>adaptation_data</i> directory contains the utterances from each speaker that can be used adapt the noise suppressor to work better for that particular speaker.
|
|
||||||
-->
|
|
||||||
|
|
||||||
### Clean Speech
|
|
||||||
* The clean speech dataset is derived from the public audio books dataset called Librivox.
|
|
||||||
* Librivox has recordings of volunteers reading over 10,000 public domain audio books in various languages, with majority of which are in English. In total, there are 11,350 speakers.
|
|
||||||
* A section of these recordings is of excellent quality, meaning that the speech was recorded using good quality microphones in a silent and less reverberant environments.
|
|
||||||
* But there are many audio recordings that are of poor speech quality with speech distortion, background noise and reverberation. Hence, it is important to filter the data based on speech quality.
|
|
||||||
* We used the online subjective test framework ITU-T P.808 to sort the book chapters by subjective quality.
|
|
||||||
* The audio chapters in Librivox are of variable length ranging from few seconds to several minutes.
|
|
||||||
* We sampled 10 random clips from each book chapter, each 10 seconds in duration. For each clip we had 3 ratings, and the Mean Opinion Score (MOS) across the all clips was used as the book chapter MOS.
|
|
||||||
* The upper quartile with respect to MOS was chosen as our clean speech dataset, which are top 25% of the clips with MOS as a metric.
|
|
||||||
* The upper quartile comprised of audio chapters with 4.3 ≤ MOS ≤ 5. We removed clips from speakers with less than 15 minutes of speech. The resulting dataset has 500 hours of speech from 2150 speakers.
|
|
||||||
* All the filtered clips are then split into segments of 31 seconds.
|
|
||||||
* Singing voice is from VocalSet corpus. It has 10.1 hrs of singing from 20 professional singers.
|
|
||||||
* Emotion speech from diverse ethnic background is provided. Emotions such as Anger, Disgust, Fear, Happy, Neutral, and Sad at four intensity levels: Low, Medium, High, Unspecified are used.
|
|
||||||
* Non-English clips consisting of tonal and non-tonal languages are included.
|
|
||||||
* More details about the clean speech data can be found in our [ICASSP 2021 DNS Challenge paper](/docs/ICASSP_2021_DNS_challenge.pdf).
|
|
||||||
<!--
|
|
||||||
FIXME: The original URL was this:
|
|
||||||
https://github.com/microsoft/DNS-Challenge/blob/icassp21/addrir/docs/ICASSP_2021_deep_noise_suppression_challenge.pdf
|
|
||||||
The branch and the file do not exist in git history. Is that the right URL?
|
|
||||||
-->
|
|
||||||
|
|
||||||
### Noise
|
|
||||||
* The noise clips were selected from Audioset and Freesound.
|
|
||||||
* Audioset is a collection of about 2 million human-labeled 10s sound clips drawn from YouTube videos and belong to about 600 audio events.
|
|
||||||
* Like the Librivox data, certain audio event classes are overrepresented. For example, there are over a million clips with audio classes music and speech and less than 200 clips for classes such as toothbrush, creak etc.
|
|
||||||
* Approximately, 42% of the clips have single class, but the rest may have 2 to 15 labels.
|
|
||||||
* Hence, we developed a sampling approach to balance the dataset in such a way that each class has at least 500 clips.
|
|
||||||
* We also used a speech activity detector (trained classifier) to remove the clips with any kind of speech activity. The reason is to avoid suppression of speech by the noise suppression model trained to suppress speech like noise.
|
|
||||||
* The resulting dataset has about 150 audio classes and 60,000 clips. We also augmented an additional 10,000 noise clips downloaded from Freesound and DEMAND databases.
|
|
||||||
* The chosen noise types are more relevant to VOIP applications.
|
|
||||||
|
|
||||||
### Room Impulse Responses (RIR)
|
|
||||||
Please use the impulse responses in the wideband dataset, as described in the [datasets/README-DNS3.md](/datasets/README-DNS3.md) file.
|
|
||||||
|
|
||||||
### Acoustic Parameters
|
|
||||||
Acoustic parameters' data is available in git at
|
|
||||||
<code>[/datasets/acoustic_params/](/datasets/acoustic_params/)</code>. Please refer to
|
|
||||||
[datasets/README.md](/datasets/README.md) for more details.
|
|
||||||
|
|
@ -35,154 +35,154 @@
|
||||||
|
|
||||||
BLOB_NAMES=(
|
BLOB_NAMES=(
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.VocalSet_48kHz_mono_000_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.VocalSet_48kHz_mono_000_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.emotional_speech_000_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.emotional_speech_000_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_000_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_000_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_001_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_001_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_002_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_002_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_003_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_003_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_004_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_004_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_005_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_005_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_006_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_006_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_007_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_007_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.french_speech_008_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.french_speech_008_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_000_0.00_3.47.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_000_0.00_3.47.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_001_3.47_3.64.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_001_3.47_3.64.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_002_3.64_3.74.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_002_3.64_3.74.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_003_3.74_3.81.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_003_3.74_3.81.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_004_3.81_3.86.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_004_3.81_3.86.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_005_3.86_3.91.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_005_3.86_3.91.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_006_3.91_3.96.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_006_3.91_3.96.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_007_3.96_4.00.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_007_3.96_4.00.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_008_4.00_4.04.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_008_4.00_4.04.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_009_4.04_4.08.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_009_4.04_4.08.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_010_4.08_4.12.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_010_4.08_4.12.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_011_4.12_4.16.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_011_4.12_4.16.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_012_4.16_4.21.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_012_4.16_4.21.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_013_4.21_4.26.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_013_4.21_4.26.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_014_4.26_4.33.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_014_4.26_4.33.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_015_4.33_4.43.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_015_4.33_4.43.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_016_4.43_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_016_4.43_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_017_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_017_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_018_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_018_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_019_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_019_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_020_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_020_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_021_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_021_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_022_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_022_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_023_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_023_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_024_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_024_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_025_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_025_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_026_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_026_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_027_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_027_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_028_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_028_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_029_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_029_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_030_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_030_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_031_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_031_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_032_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_032_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_033_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_033_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_034_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_034_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_035_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_035_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_036_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_036_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_037_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_037_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_038_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_038_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_039_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_039_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_040_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_040_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_041_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_041_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.german_speech_042_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.german_speech_042_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_000_0.00_3.98.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_000_0.00_3.98.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_001_3.98_4.21.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_001_3.98_4.21.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_002_4.21_4.40.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_002_4.21_4.40.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_003_4.40_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_003_4.40_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_004_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_004_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.italian_speech_005_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.italian_speech_005_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_000_0.00_3.75.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_000_0.00_3.75.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_001_3.75_3.88.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_001_3.75_3.88.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_002_3.88_3.96.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_002_3.88_3.96.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_003_3.96_4.02.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_003_3.96_4.02.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_004_4.02_4.06.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_004_4.02_4.06.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_005_4.06_4.10.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_005_4.06_4.10.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_006_4.10_4.13.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_006_4.10_4.13.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_007_4.13_4.16.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_007_4.13_4.16.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_008_4.16_4.19.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_008_4.16_4.19.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_009_4.19_4.21.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_009_4.19_4.21.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_010_4.21_4.24.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_010_4.21_4.24.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_011_4.24_4.26.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_011_4.24_4.26.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_012_4.26_4.29.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_012_4.26_4.29.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_013_4.29_4.31.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_013_4.29_4.31.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_014_4.31_4.33.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_014_4.31_4.33.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_015_4.33_4.35.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_015_4.33_4.35.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_016_4.35_4.38.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_016_4.35_4.38.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_017_4.38_4.40.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_017_4.38_4.40.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_018_4.40_4.42.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_018_4.40_4.42.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_019_4.42_4.45.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_019_4.42_4.45.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_020_4.45_4.48.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_020_4.45_4.48.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_021_4.48_4.52.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_021_4.48_4.52.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_022_4.52_4.57.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_022_4.52_4.57.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_023_4.57_4.67.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_023_4.57_4.67.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_024_4.67_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_024_4.67_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_025_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_025_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_026_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_026_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_027_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_027_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_028_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_028_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_029_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_029_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_030_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_030_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_031_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_031_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_032_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_032_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_033_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_033_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_034_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_034_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_035_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_035_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_036_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_036_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_037_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_037_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_038_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_038_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.read_speech_039_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.read_speech_039_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.russian_speech_000_0.00_4.31.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.russian_speech_000_0.00_4.31.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.russian_speech_001_4.31_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.russian_speech_001_4.31_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_000_0.00_4.09.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_000_0.00_4.09.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_001_4.09_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_001_4.09_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_002_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_002_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_003_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_003_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_004_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_004_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_005_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_005_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_006_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_006_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_007_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_007_NA_NA.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.spanish_speech_008_NA_NA.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.spanish_speech_008_NA_NA.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_000.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_000.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_001.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_001.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_002.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_002.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_003.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_003.tar.bz2
|
||||||
datasets_fullband/clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_004.tar.bz2
|
clean_fullband/datasets_fullband.clean_fullband.vctk_wav48_silence_trimmed_004.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_000.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_000.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_001.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_001.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_002.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_002.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_003.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_003.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_004.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_004.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_005.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_005.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.audioset_006.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.audioset_006.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_000.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.freesound_000.tar.bz2
|
||||||
datasets_fullband/noise_fullband/datasets_fullband.noise_fullband.freesound_001.tar.bz2
|
noise_fullband/datasets_fullband.noise_fullband.freesound_001.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/datasets_fullband.dev_testset_000.tar.bz2
|
datasets_fullband.dev_testset_000.tar.bz2
|
||||||
|
|
||||||
datasets_fullband/datasets_fullband.impulse_responses_000.tar.bz2
|
datasets_fullband.impulse_responses_000.tar.bz2
|
||||||
)
|
)
|
||||||
|
|
||||||
###############################################################
|
###############################################################
|
||||||
|
|
||||||
AZURE_URL="https://dns4public.blob.core.windows.net/dns4archive"
|
AZURE_URL="https://dns4public.blob.core.windows.net/dns4archive/datasets_fullband"
|
||||||
|
|
||||||
OUTPUT_PATH="."
|
OUTPUT_PATH="./datasets_fullband"
|
||||||
|
|
||||||
mkdir -p $OUTPUT_PATH/datasets_fullband/{clean_fullband,noise_fullband}
|
mkdir -p $OUTPUT_PATH/{clean_fullband,noise_fullband}
|
||||||
|
|
||||||
for BLOB in ${BLOB_NAMES[@]}
|
for BLOB in ${BLOB_NAMES[@]}
|
||||||
do
|
do
|
||||||
|
|
|
||||||
Загрузка…
Ссылка в новой задаче