Noise robust speaker verification via the fusion of SNR-independent and SNR-dependent PLDA

Pang, Xiaomin; Mak, Man‐Wai

doi:10.1007/s10772-015-9310-8

Cited by 15 publications

(1 citation statement)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These techniques are based on single features and do not capture the full range of information present in a speaker's voice, which can affect the accuracy of identification. 2,4 Speaker pattern classification is the process of classifying speakers based on their voice characteristic parameters. This is done by building a speaker pattern using techniques such as Support Vector Machines (SVMs), neural networks, Hidden Markov Patterns (HMMs), and vector quantization patterns.…”

Section: Related Workmentioning

confidence: 99%

Deep Learning and Limited Boltzmann Machines for Speaker Recognition

P.Mathiyalagan

2023

IJRITCC

View full text Add to dashboard Cite

Speaker recognition has become an essential aspect of modern voice-based systems such as security and authentication applications. In this research, we propose a new method for speaker recognition based on deep learning and limited Boltzmann machines. The method comprises preemphasis and overlapping type framing, endpoint detection, feature extraction, and training of a depth belief network pattern using a limited Boltzmann machine layer. The Softmax graders are added in the top layer of the pattern, and the speaker's phonetic feature is input into the pattern for training. The likelihood probability of other speakers' phonetic features is calculated, and the speaker corresponding to the maximum probability is identified as the recognized result. The results show that the proposed method outperforms other state-of-the-art methods, achieving high accuracy and robustness to noise and signal variations.

show abstract

Section: Related Workmentioning

confidence: 99%

Deep Learning and Limited Boltzmann Machines for Speaker Recognition

P.Mathiyalagan

2023

IJRITCC

View full text Add to dashboard Cite

show abstract

Robust Threshold Selection for Environment Specific Voice in Speaker Recognition

Kanrar

2022

Wireless Pers Commun

View full text Add to dashboard Cite

Machine Learning for Speaker Recognition

Mak

Chien

2020

View full text Add to dashboard Cite

This chapter defines the term "speaker recognition" and looks at this technology from a high-level perspective. Then, it introduces the fundamental concepts of speaker recognition, including feature extraction, speaker modeling, scoring, and performance measures. Fundamentals of Speaker RecognitionFrom time to time we hear that the information of millions of customers of remote services has been compromised. These security leaks cause concerns about the security of the remote services that everyone uses on a daily basis. While these remote services bring convenience and benefit to users, they are also gold mines for criminals to carry out fraudulent acts. The conventional approach to user authentication, such as usernames and passwords, is no longer adequate for securing these services. A number of companies have now introduced voice biometrics as a complement to the conventional username-password approach. With this new authentication method, it is much harder for the criminals to imitate the legitimate users. Voice biometrics can also reduce the risk of leaking customers' information caused by social engineering fraudulence. Central to voice biometrics authentication is speaker recognition. Another application domain of voice biometrics is to address the privacy issues of smartphones, home assistants, and smart speakers. With the increasing intelligence capabilities of these devices, we can interact with them as if they were human. Because these devices are typically used solely by their owners or their family members and speech is the primary means of interaction, it is natural to use the voice of the owners for authentication, i.e., a device can only be used by its owner.Speaker recognition is a technique to recognize the identity of a speaker from a speech utterance. As shown in Figure 1.1, in terms of recognition tasks, speaker recognition can be categorized into speaker identification, speaker verification, and speaker diarization. In all of these tasks, the number of speakers involved can be fixed (closed set) or varied (open set).Speaker identification is to determine whether the voice of an unknown speaker matches one of the N speakers in a dataset, where N could be very large (thousands). It is a one-to-many mapping and it is often assumed that the unknown voice

show abstract

Noise robust speaker verification via the fusion of SNR-independent and SNR-dependent PLDA

Cited by 15 publications

References 32 publications

Deep Learning and Limited Boltzmann Machines for Speaker Recognition

Deep Learning and Limited Boltzmann Machines for Speaker Recognition

Robust Threshold Selection for Environment Specific Voice in Speaker Recognition

Machine Learning for Speaker Recognition

Contact Info

Product

Resources

About