Vocal Tract Length Perturbation (VTLP) improves speech recognition

By Navdeep Jaitly, Geoffrey E. Hinton. 2013

448 citations recorded by the source corpus.

How researchers see this paper

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

In Speech & Audio

Citations in the corpus

SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition (2019, methodology): “Vocal Tract Length Normalization has been adapted for data augmentation in [11].”
Deep Speech 2 : End-to-End Speech Recognition in English and Mandarin (2015, methodology): “. Techniques used for data augmentation in speech range from simple noise addition [26] to complex perturbations such as simulating changes to the vocal tract length and rate of speech of the speaker [31, 35]. Existing speech systems can also be used to bootstrap new data collection. In one approach”
Data Recombination for Neural Semantic Parsing (2016, neutral): “These techniques have proven effective in areas like computer vision (Krizhevsky et al., 2012) and speech recognition (Jaitly and Hinton, 2013).”
Specaugment on Large Scale Datasets (2019, methodology): “Vocal Tract Length Perturbation has been introduced in the context of data augmentation for ASR in [8], and explored further in [9].”
Learning Latent Representations for Speech Generation and Transformation (2017, background): “Understanding the influence of these factors on the speech signal is an important problem, which can be used for a wide variety of applications, including, but not limited to adaptation and data augmentation for speech recognition [1, 2], voice conversion [3, 4, 5], and speech compression [6].”
Conditional BERT Contextual Augmentation (2018, background): “Recent years have witnessed great success in applying data augmentation in the ﬁeld of speech area(Jaitly and Hinton, 2013; Ko et al., 2015) and computer vision(Krizhevsky et al., 2012; Simard et al., 1998; Szegedy et al., 2015).”
An empirical survey of data augmentation for time series classification with neural networks (2020, methodology): “Vocal Tract Length Perturbation (VTLP) [22], for example, is an extension to Vocal Tract Length Normalization (VTLN) [57] that adds variability instead of removes it.”
Data Augmentation for deep neural network acoustic modeling (2014, methodology): “VTLP was first proposed in [3] with experiments on TIMIT database where for each utterance in the training set a warping factor α is randomly chosen from [0.”
Semi-supervised ASR by End-to-end Self-training (2020, background): “And various data augmentation techniques have demonstrated consistent improvement for ASR [10, 11, 12, 13].”
Data augmentation for low resource languages (2014, methodology): “These include standard statistical approaches, such as maximum a posteriori (MAP) estimation [6], and data augmentation [7, 8, 9, 10, 11, 12].”
Convolutional Recurrent Neural Network and Data Augmentation for Audio Tagging with Noisy Labels and Minimal Supervision (2019, result): “It is worth noting that the frequency warping performed here is very similar to Vocal Tract Length Pertubation [14].”
Deep Learning for Audio Signal Processing (2019, methodology): “On a historical note, in ASR, MIR, and environmental sound analysis, deep models have replaced support vector machines for sequence classification, and GMM-HMMs for sequence transduction.”
Dreaming More Data: Class-dependent Distributions over Diffeomorphisms for Learned Data Augmentation (2015, methodology): “In work on speech recognition, Jaitly and Hinton [27] apply Vocal Tract Normalization (VTLN) as a way to artiﬁcially transform utterances of one speaker to the voice of another.”
Can Speaker Augmentation Improve Multi-Speaker End-to-End TTS? (2020, background): “We create a large mixed dataset for TTS training using both VCTK and a variety of corpora that were collected for ASR and thus contain a variety of different recording conditions and English dialects.”
Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relations (2018, unknown): “, 2015) and speech (Jaitly and Hinton, 2015; Ko et al., 2015).”
A Survey on Deep Learning for Data-Driven Soft Sensors (2021, background): “Although, it may be not easy to collect large amount of data from real scenarios, creating new fake data is meaningful for some speciﬁc tasks, such as object recognition [68]andspeechrecognition[69].”