Small Energy Masking for Improved Neural Network Training for End-To-End Speech Recognition

Kim, Chanwoo; Kim, Kwangyoun; Indurthi, Sathish Reddy

doi:10.1109/icassp40776.2020.9054382

Cited by 8 publications

(4 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This can be done with special data augmentation. For instance in speech recognition, it can be done by randomly removing low-energy parts of the recording [45]. In our case, it could be done by randomly inserting bias into images during the training, similar to online data augmentation.…”

Section: Discussionmentioning

confidence: 99%

Towards Explainable Classifiers Using the Counterfactual Approach - Global Explanations for Discovering Bias in Data

Kwasigroch

2020

Journal of Artificial Intelligence and Soft Computing Research

View full text Add to dashboard Cite

The paper proposes summarized attribution-based post-hoc explanations for the detection and identification of bias in data. A global explanation is proposed, and a step-by-step framework on how to detect and test bias is introduced. Since removing unwanted bias is often a complicated and tremendous task, it is automatically inserted, instead. Then, the bias is evaluated with the proposed counterfactual approach. The obtained results are validated on a sample skin lesion dataset. Using the proposed method, a number of possible bias-causing artifacts are successfully identified and confirmed in dermoscopy images. In particular, it is confirmed that black frames have a strong influence on Convolutional Neural Network’s prediction: 22% of them changed the prediction from benign to malignant.

show abstract

Section: Discussionmentioning

confidence: 99%

Towards Explainable Classifiers Using the Counterfactual Approach - Global Explanations for Discovering Bias in Data

Kwasigroch

2020

Journal of Artificial Intelligence and Soft Computing Research

View full text Add to dashboard Cite

show abstract

“…Further, 7th root compression is applied instead of logarithmic, as in Gammatone or PLP features [25,26]. The alternative feature extraction pipeline is also extended with small-energy masking (SEM) perturbation [27]. Relative to the peak energy in a given utterance, the method masks time-frequency bins with small energy in the Mel-spectral domain.…”

Section: Methodsmentioning

confidence: 99%

On the Limit of English Conversational Speech Recognition

2021

View full text Add to dashboard Cite

In our previous work we demonstrated that a single headed attention encoder-decoder model is able to reach state-of-theart results in conversational speech recognition. In this paper, we further improve the results for both Switchboard 300 and 2000. Through use of an improved optimizer, speaker vector embeddings, and alternative speech representations we reduce the recognition errors of our LSTM system on Switchboard-300 by 4% relative. Compensation of the decoder model with the probability ratio approach allows more efficient integration of an external language model, and we report 5.9% and 11.5% WER on the SWB and CHM parts of Hub5'00 with very simple LSTM models. Our study also considers the recently proposed conformer, and more advanced self-attention based language models. Overall, the conformer shows similar performance to the LSTM; nevertheless, their combination and decoding with an improved LM reaches a new record on Switchboard-300, 5.0% and 10.0% WER on SWB and CHM. Our findings are also confirmed on Switchboard-2000, and a new state of the art is reported, practically reaching the limit of the benchmark.

show abstract

“…Room impulse response simulation and adding point-source noises were proposed for far-field ASR [25]. Inspired by input dropout, [26] proposed to improve the noise robustness of CNN acoustic models by discarding input features, and [27] proposed to mask time-frequency bins with energy lower than randomised thresholds. Recently, SpecAugment [28] (SA) was proposed to augment speech data by warping spectrograms along the time axis, and masking time and/or frequency bands in the spectral domain.…”

Section: Related Workmentioning

confidence: 99%

Auditory-Based Data Augmentation for End-to-End Automatic Speech Recognition

Tu¹,

Deadman²,

Ma³

et al. 2022

Preprint

View full text Add to dashboard Cite

End-to-end models have achieved significant improvement on automatic speech recognition. One common method to improve performance of these models is expanding the data-space through data augmentation. Meanwhile, human auditory inspired front-ends have also demonstrated improvement for automatic speech recognisers. In this work, a well-verified auditory-based model, which can simulate various hearing abilities, is investigated for the purpose of data augmentation for end-to-end speech recognition. By introducing the auditory model into the data augmentation process, end-to-end systems are encouraged to ignore variation from the signal that cannot be heard and thereby focus on robust features for speech recognition. Two mechanisms in the auditory model, spectral smearing and loudness recruitment, are studied on the LibriSpeech dataset with a transformer-based end-to-end model. The results show that the proposed augmentation methods can bring statistically significant improvement on the performance of the state-of-the-art SpecAugment.

show abstract

Small Energy Masking for Improved Neural Network Training for End-To-End Speech Recognition

Cited by 8 publications

References 17 publications

Towards Explainable Classifiers Using the Counterfactual Approach - Global Explanations for Discovering Bias in Data

Towards Explainable Classifiers Using the Counterfactual Approach - Global Explanations for Discovering Bias in Data

On the Limit of English Conversational Speech Recognition

Auditory-Based Data Augmentation for End-to-End Automatic Speech Recognition

Contact Info

Product

Resources

About