Elastic spectral distortion for low resource speech recognition with deep neural networks

By Naoyuki Kanda, Ryu Takeda, Yasunari Obuchi. 2013

129 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing, 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

1. Jinyu Li
2. Shinji Watanabe
3. Yu Zhang
4. Quoc V. Le
5. Dong Yu
6. Gabriel Synnaeve
7. Tara N. Sainath
8. Yoshua Bengio

In Natural Language Processing

1. Shinji Watanabe
2. Yu Zhang
3. Tara N. Sainath
4. Jinyu Li
5. William Chan
6. Ralf Schlüter
7. Yonghui Wu
8. Quoc V. Le

In Signal Processing

1. Shinji Watanabe
2. Jinyu Li
3. Yu Zhang
4. Tara N. Sainath
5. Dong Yu
6. Takaaki Hori
7. Gabriel Synnaeve
8. Yonghui Wu

Citations in the corpus

• Randaugment: Practical automated data augmentation with a reduced search space (2019, methodology): “Data augmentation is a widely used method for generating additional data to improve machine learning systems, for image classification [43, 23, 7, 54], object detection [13], instance segmentation [10], and speech recognition [21, 16, 36].”
• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition (2019, background): “For example, in [9, 10], artificial data was augmented for low resource speech recognition tasks.”
• Specaugment on Large Scale Datasets (2019, background): “Artificial data augmentation for low resource speech recognition tasks has been studied in [6, 7].”
• Data Augmenting Contrastive Learning of Speech Representations in the Time Domain (2020, methodology): “Data augmentation is a common method to improve supervised training in speech applications, especially in low resource settings [25, 26].”
• 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].”
• End-to-End Speech Recognition: A Survey (2023, methodology): “Most data augmentation methods perform data perturbation by exploiting certain dimensions of speech signal variation: speed perturbation [201], [202], vocal tract length perturbation[201],[203],frequencyaxisdistortion[201],sequencenoise injection[204],SpecAugment[205],orsemanticmask[206].”
• Data Augmentation for Deep Neural Network Acoustic Modeling (2015, methodology): “…using label-preserving transforma­tions has been widely used in neural network based pattern recognition in computer vision and image recognition tasks [11] [13] where transformations such as translation, defor­mation and re.ection have led to signi.cant improvements in recognition accuracy.”
• Deep Learning for Audio Signal Processing (2019, methodology): “For ASR, [77] and [78] independently proposed to transform speech excerpts by pitch”
• Acoustic Modeling for Distant Multi-talker Speech Recognition with Single- and Multi-channel Branches (2019, background): “One way to improve distant multi-talker speech recognition is to improve the robustness of acoustic models (AMs) with data augmentation techniques [10], better training objectives [7, 11–14], improved model architecture [15, 16], etc.”
• Speech recognition and keyword spotting for low-resource languages: Babel project research at CUED (2014, methodology): “One technique is to artificially create more data with known transcriptions, for example using acoustic data perturbation [12, 13, 14] or speech synthesis [15].”
• Lattice-free State-level Minimum Bayes Risk Training of Acoustic Models (2018, methodology): “In an earlier version of DNN-based ASR, AMs were trained using a frame-level cross-entropy (CE) loss criterion [1, 2, 3], which provided signifi-cant accuracy improvements compared to the Gaussian mixture model (GMM)-based AMs [3, 4, 5].”
• RandAugment: Practical data augmentation with no separate search (2019, methodology): “Data augmentation is a widely used method for generating additional data to improve machine learning systems, for image classification [42, 22, 7, 53], object detection [13], instance segmentation [10], and speech recognition [20, 15, 35].”
• Audio augmentation for speech recognition (2015, methodology): “In [5, 6] the use of data augmentation on low resource languages, where the amount of training data is comparatively small (∼ 10 hrs), was investigated.”
• An analysis of environment, microphone and data simulation mismatches in robust speech recognition (2017, unknown): “Fortunately, this case against simulation does not arise for all techniques: most modern enhancement and ASR techniques can benefit from data augmentation and simulation (Kanda et al., 2013; Brutti and Matassoni, 2016).”
• Maximum a posteriori Based Decoding for CTC Acoustic Models (2016, background): “The hybrid framework of hidden Markov models (HMMs) and deep neural networks (DNNs) has achieved great progress in speech recognition [1, 2], showing high recognition accuracies in numerous scenarios [3, 4, 5].”