FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition
As the architecture of deep learning-based speech recognizers has recently changed to the end-to-end style, increasing the effective amount of training data has become an important issue. To tackle this issue, various data augmentation techniques to create additional training data by transforming la...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-07-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/12/15/7619 |
| _version_ | 1827626127831597056 |
|---|---|
| author | Seong-Su Lim Oh-Wook Kwon |
| author_facet | Seong-Su Lim Oh-Wook Kwon |
| author_sort | Seong-Su Lim |
| collection | DOAJ |
| description | As the architecture of deep learning-based speech recognizers has recently changed to the end-to-end style, increasing the effective amount of training data has become an important issue. To tackle this issue, various data augmentation techniques to create additional training data by transforming labeled data have been studied. We propose a method called FrameAugment to augment data by changing the speed of speech locally for selected sections, which is different from the conventional speed perturbation technique that changes the speed of speech uniformly for the entire utterance. To change the speed of the selected sections of speech, the number of frames for the randomly selected sections is adjusted through linear interpolation in the spectrogram domain. The proposed method is shown to achieve 6.8% better performance than the baseline in the WSJ database and 9.5% better than the baseline in the LibriSpeech database. It is also confirmed that the proposed method further improves speech recognition performance when it is combined with the previous data augmentation techniques. |
| first_indexed | 2024-03-09T12:48:29Z |
| format | Article |
| id | doaj.art-e82daf3d55f945edb122b550b162e9be |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-09T12:48:29Z |
| publishDate | 2022-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-e82daf3d55f945edb122b550b162e9be2023-11-30T22:10:05ZengMDPI AGApplied Sciences2076-34172022-07-011215761910.3390/app12157619FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech RecognitionSeong-Su Lim0Oh-Wook Kwon1Major in Control and Robot Engineering, Chungbuk National University, Cheongju 28644, KoreaDepartment of Intelligent Systems and Robotics, Chungbuk National University, Cheongju 28644, KoreaAs the architecture of deep learning-based speech recognizers has recently changed to the end-to-end style, increasing the effective amount of training data has become an important issue. To tackle this issue, various data augmentation techniques to create additional training data by transforming labeled data have been studied. We propose a method called FrameAugment to augment data by changing the speed of speech locally for selected sections, which is different from the conventional speed perturbation technique that changes the speed of speech uniformly for the entire utterance. To change the speed of the selected sections of speech, the number of frames for the randomly selected sections is adjusted through linear interpolation in the spectrogram domain. The proposed method is shown to achieve 6.8% better performance than the baseline in the WSJ database and 9.5% better than the baseline in the LibriSpeech database. It is also confirmed that the proposed method further improves speech recognition performance when it is combined with the previous data augmentation techniques.https://www.mdpi.com/2076-3417/12/15/7619data augmentationend-to-end speech recognitionframe rate |
| spellingShingle | Seong-Su Lim Oh-Wook Kwon FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition Applied Sciences data augmentation end-to-end speech recognition frame rate |
| title | FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition |
| title_full | FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition |
| title_fullStr | FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition |
| title_full_unstemmed | FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition |
| title_short | FrameAugment: A Simple Data Augmentation Method for Encoder–Decoder Speech Recognition |
| title_sort | frameaugment a simple data augmentation method for encoder decoder speech recognition |
| topic | data augmentation end-to-end speech recognition frame rate |
| url | https://www.mdpi.com/2076-3417/12/15/7619 |
| work_keys_str_mv | AT seongsulim frameaugmentasimpledataaugmentationmethodforencoderdecoderspeechrecognition AT ohwookkwon frameaugmentasimpledataaugmentationmethodforencoderdecoderspeechrecognition |