Speech Recognition using EfficientNet

Lu, Qidong; Li, Yingying; Qin, Zhiliang; Liu, Xiaowei; Xie, Yun

doi:10.1145/3404716.3404717

Cited by 13 publications

(5 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We investigate 14 modern DNN architectures: seven variations of EfficientNetV2 [9] and seven variations of Efficient-NetV1 [8]. All these architectures were originally designed for image classification and have since been applied to various other problems such as speech recognition [26]. The accuracy of these architectures on the ImageNet dataset [27] ranges from 77.1% to 85.7%.…”

Section: Resultsmentioning

confidence: 99%

Dynamic Split Computing for Efficient Deep Edge Intelligence

Bakhtiarnia¹,

Milošević²,

Zhang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Deploying deep neural networks (DNNs) on IoT and mobile devices is a challenging task due to their limited computational resources. Thus, demanding tasks are often entirely offloaded to edge servers which can accelerate inference, however, it also causes communication cost and evokes privacy concerns. In addition, this approach leaves the computational capacity of end devices unused. Split computing is a paradigm where a DNN is split into two sections; the first section is executed on the end device, and the output is transmitted to the edge server where the final section is executed. Here, we introduce dynamic split computing, where the optimal split location is dynamically selected based on the state of the communication channel. By using natural bottlenecks that already exist in modern DNN architectures, dynamic split computing avoids retraining and hyperparameter optimization, and does not have any negative impact on the final accuracy of DNNs. Through extensive experiments, we show that dynamic split computing achieves faster inference in edge computing environments where the data rate and server load vary over time.

show abstract

Section: Resultsmentioning

confidence: 99%

Dynamic Split Computing for Efficient Deep Edge Intelligence

Bakhtiarnia¹,

Milošević²,

Zhang³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…[21] distorted the vocal tract length (VTLP) managing to improve ASR models by 2.5% TER (Token Error Rate). [22] made a substantial improvement in this technique by distorting noise addition, velocity adjustment and pitch shifting in the original audios, managing to reduce WER by 5.1%. DA techniques using a TTS model are also delivering good results.…”

Section: Data Augmentationmentioning

confidence: 99%

“…As a second baseline, the method of DA through audio distortion proposed by [22] was developed to augment the speech data of the original corpus. The nlpaug13 library was used to manipulate the training audios (99 hours) by modifying the speed according to a randomly selected coefficient in the range between 0.85 and 1.15, which is where this DA technique performs best.…”

Section: Distortion Baselinementioning

confidence: 99%

Data Augmentation for Low-Resource Quechua ASR Improvement

Zevallos¹,

Bel²,

Cámbara³

et al. 2022

Interspeech 2022

View full text Add to dashboard Cite

Automatic Speech Recognition (ASR) is a key element in new services that helps users to interact with an automated system. Deep learning methods have made it possible to deploy systems with word error rates below 5% for ASR of English. However, the use of these methods is only available for languages with hundreds or thousands of hours of audio and their corresponding transcriptions. For the so-called low-resource languages to speed up the availability of resources that can improve the performance of their ASR systems, methods of creating new resources on the basis of existing ones are being investigated. In this paper we describe our data augmentation approach to improve the results of ASR models for low-resource and agglutinative languages. We carry out experiments developing an ASR for Quechua using the wav2letter++ model. We reduced WER by 8.73% through our approach to the base model. The resulting ASR model obtained 22.75% WER and was trained with 99 hours of original resources and 99 hours of synthetic data obtained with a combination of text augmentation and synthetic speech generation.

show abstract

“…This model could be widely applicable for transfer learning in sound classification. Lu et al used the morphology of spectrograms as the input pattern to recognize speech using an EfficientNet model [10]. Padi et al employed transfer learning to improve the accuracy of speech emotion recognition through spectrogram augmentation [7].…”

Section: Introductionmentioning

confidence: 99%

A Wearable Assistant Device for the Hearing Impaired to Recognize Emergency Vehicle Sirens with Edge Computing

Chin,

Lin,

Wang

et al. 2023

Sensors

View full text Add to dashboard Cite

Wearable assistant devices play an important role in daily life for people with disabilities. Those who have hearing impairments may face dangers while walking or driving on the road. The major danger is their inability to hear warning sounds from cars or ambulances. Thus, the aim of this study is to develop a wearable assistant device with edge computing, allowing the hearing impaired to recognize the warning sounds from vehicles on the road. An EfficientNet-based, fuzzy rank-based ensemble model was proposed to classify seven audio sounds, and it was embedded in an Arduino Nano 33 BLE Sense development board. The audio files were obtained from the CREMA-D dataset and the Large-Scale Audio dataset of emergency vehicle sirens on the road, with a total number of 8756 files. The seven audio sounds included four vocalizations and three sirens. The audio signal was converted into a spectrogram by using the short-time Fourier transform for feature extraction. When one of the three sirens was detected, the wearable assistant device presented alarms by vibrating and displaying messages on the OLED panel. The performances of the EfficientNet-based, fuzzy rank-based ensemble model in offline computing achieved an accuracy of 97.1%, precision of 97.79%, sensitivity of 96.8%, and specificity of 97.04%. In edge computing, the results comprised an accuracy of 95.2%, precision of 93.2%, sensitivity of 95.3%, and specificity of 95.1%. Thus, the proposed wearable assistant device has the potential benefit of helping the hearing impaired to avoid traffic accidents.

show abstract

Speech Recognition using EfficientNet

Cited by 13 publications

References 18 publications

Dynamic Split Computing for Efficient Deep Edge Intelligence

Dynamic Split Computing for Efficient Deep Edge Intelligence

Data Augmentation for Low-Resource Quechua ASR Improvement

A Wearable Assistant Device for the Hearing Impaired to Recognize Emergency Vehicle Sirens with Edge Computing

Contact Info

Product

Resources

About