ASVspoof 2021 Baseline Systems

Last update: Dec 28, 2022

Related tags

Miscellaneous 2021

Overview

ASVspoof 2021 Baseline Systems

Baseline systems are grouped by task:

Speech Deepfake (DF)
Logical Access (LA)
Physical Access (PA)

Please find more details in the evaluation plan.

Reference code is provided for the following systems:

Baseline-CQCC-GMM (Matlab & Python)
CQCC feature extraction with GMM classifier
Baseline-LFCC-GMM (Matlab & Python)
LFCC feature extraction with GMM classifier
Baseline-LFCC-LCNN (PyTorch)
LFCC feature extraction with LCNN classifier (DNN)
Baseline-RawNet2 (PyTorch)
End-to-End DNN classifier

Comments

How to download the ground-truth labels of the evaluation sets of this year

I cannot access the keys, but something like this: version https://git-lfs.github.com/spec/v1 oid sha256:466a73bff42ac654626e784d35ee990f596872e65705769a112e140c8a4d71b9 size 21255945

opened by changtaoli 3
> Missed parameters of fs, fmin, fmax in extract_cqccc

Missed parameters of fs, fmin, fmax in extract_cqccc

https://github.com/asvspoof-challenge/2021/blob/17c3ce9759f532f2c76d5017d33ac3321142d9c7/LA/Baseline-CQCC-GMM/python/gmm.py#L72

After fixing this error, string indices must be integers error occurs in

https://github.com/asvspoof-challenge/2021/blob/17c3ce9759f532f2c76d5017d33ac3321142d9c7/LA/Baseline-CQCC-GMM/python/gmm.py#L43

Hello, has your problem been solved? I also encountered this problem.

Originally posted by @yuant-gif in https://github.com/asvspoof-challenge/2021/issues/2#issuecomment-1006567437

opened by veesa4826 1
Demo for cqcc not working

There is a error in line 107 for the function resample in CQCC toolbox. The error is Check for incorrect argument data type or missing argument in call to function 'resample'.

https://github.com/asvspoof-challenge/2021/blob/80f8d4c8505a2b8ad94a76dbade72fc16217e689/LA/Baseline-CQCC-GMM/matlab/CQCC/cqcc.m#L107

opened by rishabh004-ai 0
Unable to download ASVSpoof Challenge dataset 2021

Hi, For the last 2 days, I have been trying to download the dataset - LA.zip from both : -Zenodo site ( https://zenodo.org/record/4837263#.YiHDfHNBy5c ) -University of Edinburgh data share ( https://datashare.ed.ac.uk/handle/10283/3336 )

The download is extremely slow ( in spite of my broadband download speed of 30Mbps ) and gets cancelled mid way.

Could you let me know if there are any other servers that this data is present on and can be downloaded quickly.

Thanks in advance.

opened by PS-AI 2
Pre-trained model for PA task

Hello,

Seems the pre-trained model provided for PA is the one that is trained for LA task. Could you please point me to a link for downloading pre-trained RawNet2 for PA please?

Thanks, Ali

opened by AliShahin 0

Dataset_ASVspoof2021_eval 返回的为什么是x_inp, key 而不是返回x_inp, label[key]

class Dataset_ASVspoof2021_eval(Dataset):
    def __init__(self, list_IDs, base_dir):
        """
            self.list_IDs	: list of strings (each string: utt key),
        """

        self.list_IDs = list_IDs
        self.base_dir = base_dir

    def __len__(self):
        return len(self.list_IDs)

    def __getitem__(self, index):
        self.cut = 64600  # take ~4 sec audio (64600 samples)
        key = self.list_IDs[index]
        X, fs = librosa.load(self.base_dir + 'flac/' + key + '.flac', sr=16000)
        X_pad = pad(X, self.cut)
        x_inp = Tensor(X_pad)
        return x_inp, key

opened by Liuyang0001 4

The loss function of "LA/Baseline-RawNet2"

As i read from "the read document of pytorch", the "CrossEntropyLoss() function" is criterion combintion of nn.LogSoftmax() and nn.NLLLoss(). Why there is still a nn.LogSoftmax() layer in the last layer of RawNet() and if i remove the nn.LogSoftmax() there would be Bad results.

opened by Amforever 3

Owner

GitHub

Implent of Oracle Base line and Lea-3 Baseline

Oracle-Baseline Implent of Oracle Base line and Lea-3 Baseline Oracle Oracle : This model is used to obtain an oracle with a greedy algorithm similar

2 Nov 12, 2021

Structural basis for solubility in protein expression systems

Structural basis for solubility in protein expression systems Large-scale protein production for biotechnology and biopharmaceutical applications rely

16 Aug 18, 2022

LSO, also known as Linux Swap Operator, is a software with both GUI and terminal versions that you can manage the Swap area for Linux operating systems.

LSO - Linux Swap Operator Türkçe - LSO Nedir? LSO, diğer adıyla Linux Swap Operator Linux işletim sistemleri için Swap alanını yönetebileceğiniz hem G

4 Feb 9, 2022

Control System Packer is a lightweight, low-level program to transform energy equations into the compact libraries for control systems.

Control System Packer is a lightweight, low-level program to transform energy equations into the compact libraries for control systems. Packer supports Python ?? , C ?? and C++ ?? libraries.