Language Recognition via i-vectors and Dimensionality Reduction

By Najim Dehak, Pedro A. Torres-Carrasquillo, Douglas A. Reynolds, Réda Dehak. 2011

481 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Representation & Self-Supervised Learning

In Speech & Audio

In Signal Processing

Citations in the corpus

The People's Speech: A Large-Scale Diverse English Speech Recognition Dataset for Commercial Usage (2021, background): “We think locality-sensitive hashing [21] of ivectors [24], xvectors [30], or other text-independent speaker embedding is a promising approach to find speaker overlap in utterances.”
Exploring the Encoding Layer and Loss Function in End-to-End Speaker and Language Recognition System (2018, background): “Inrecentdecades, the classical GMM i-vector approach and its variants have dominated multiple kinds of paralinguistic speech attribute recognition ﬁelds for its superior performance, simplicity and efﬁciency [3, 4].”
A unified deep neural network for speaker and language recognition (2015, methodology): “Applications of the indirect method have used a DNN trained for ASR to extract frame-level features [2, 3, 10], accumulate a multinomial vector [7] or accumulate multi-modal statistics [6, 8] that were then used to train an i-vector system [11, 12].”
Feature Representation of Short Utterances Based on Knowledge Distillation for Spoken Language Identification (2018, background): “I-vector techniques with conventional classifiers, such as, support vector machine (SVM), probabilistic linear discriminant analysis (PLDA), and deep neural network (DNN), have demonstrated their effectiveness and obtained state-of-the-art performance in many systems, especially on relative longer u”
Network Adaptation Strategies for Learning New Classes without Forgetting the Original Ones (2019, background): “Each speech segment is represented by an ivector of 400 components [14].”
LID-Senones and Their Statistics for Language Identification (2018, methodology): “LID-net evaluated in this paper uses a SPP pooling size of [1, 1] since this was found to give consistently better performance than [1, 2] SPP.”
Deep Neural Network Approaches to Speaker and Language Recognition (2015, methodology): “Gaussian vector [7], [11] or accumulate multi-modal statistics [6], [8] that were then used to train an i-vector system [13], [14].”
A Semisupervised Approach for Language Identification based on Ladder Networks (2016, background): “Each speech segment is represented by an i-vector of 400 components [17].”
Can Speaker Augmentation Improve Multi-Speaker End-to-End TTS? (2020, background): “Dialect identification can be considered as a subtask of spoken language recognition, and in general, approaches from speaker recognition tasks can be directly transferred to dialect identification, see [24, 25, 26, 27].”
Speaker verification and spoken language identification using a generalized i-vector framework with phonetic tokenizations and tandem features (2014, methodology): “Second, in order to reduce the dimensionality of the concatenated statistics vectors, a single factor analysis is adopt to generate a low dimensional total variability space which jointly models language, speaker and channel variabilities all together [1].”
Spoken Language Recognition: From Fundamentals to Practice (2013, methodology): “Most recently, another factor analysis technique, i-vector [34], has become very popular in speaker recognition and has been also introduced to language recognition [35].”
Generalized I-vector Representation with Phonetic Tokenizations and Tandem Features for both Text Independent and Text Dependent Speaker Verification (2015, methodology): “Second, in order to reduce the high dimension of the concatenated statistics supervectors, a single factor analysis is adopt to generate a low dimensional total variability space which jointly models language, speaker and channel variabilities all together [4].”
Knowledge Distillation-Based Representation Learning for Short-Utterance Spoken Language Identification (2020, methodology): “The i-vector is a ﬁxed-length low-dimensional representation of acoustic variations, and it has been successfully used in many systems, such as language identiﬁcation and speaker recognition tasks [3]–[11].”
Automatic language identification using deep neural networks (2014, background): “In particular, guided by the advances on speaker verification, the use of i-vector extractors as a front-end followed by diverse classification mechanisms has become the state-of-the-art in acoustic LID systems [8][9].”
Spoken Language Identification using ConvNets (2019, unknown): “These approaches required significant domain knowledge [4,16].”
Interactive Learning of Teacher-student Model for Short Utterance Spoken Language Identification (2019, background): “I-vector approaches have been demonstrated their effectiveness and obtained state-of-the-art performance in many LID tasks, especially on relatively longer utterance tasks [3, 4, 5, 6, 7, 8].”