ABSTRACT
The publication of the Whisper model by OpenAI inspired us with the idea of a web platform that provides voice-to-text conversion services for Vietnamese people. Using Whisper’s powerful generalization capabilities, we have developed a web application with three main features: record-to-text, file-to-text, and subtitles generator for YouTube. We first fine-tuned Whisper with our target language dataset then deployed the model as a Rest API using the Python Flask framework with three paths for three different tasks. The web application has been developed using ReactJS, a popular JavaScript library for building user interfaces. Its architecture is grounded in component-based design principles, which means that the application is structured into reusable and modular components, enhancing code maintainability and scalability. The web application has been developed using ReactJS, a popular JavaScript library for building user interfaces. Its architecture is grounded in component-based design principles, which means that the application is structured into reusable and modular components, enhancing code maintainability and scalability. The record-to-text function will allow users to record audio on the web page, and then the audio will be processed and converted to text. As for the file-to-text function, the website will receive audio files uploaded by users and will return the transcript text of that audio file. And finally the subtitles generator for YouTube function, where users can enter the YouTube link as input, wait for the website to process and the website will display that video with the transcript attached to the video based on the timestamps of each transcript. This project can inspire and encourage the testing and application of new automatic speech recognition (ASR) models in specific applications.
- Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. 2020. wav2vec 2.0: A framework for self-supervised learning of speech representations. Advances in neural information processing systems 33 (2020), 12449–12460.Google Scholar
- James Baker. 1975. The DRAGON system–An overview. IEEE Transactions on Acoustics, speech, and signal Processing 23, 1 (1975), 24–29.Google ScholarCross Ref
- William C Dersch. [n. d.]. IBM Archives: IBM Shoebox. URL http://www-03. ibm. com/ibm/history/exhibits/specialprod1/specialprod1 { _} 7 ([n. d.]).Google Scholar
- Anmol Gulati, James Qin, Chung-Cheng Chiu, and Niki Parmar. 2020. Conformer: Convolution-augmented transformer for speech recognition. arXiv preprint arXiv:2005.08100 (2020).Google Scholar
- Anmol Gulati, James Qin, Chung-Cheng Chiu, Niki Parmar, and Zhang. 2020. Conformer: Convolution-augmented transformer for speech recognition. arXiv preprint arXiv:2005.08100 (2020).Google Scholar
- DV Hai and ASR Challenge. 2021. Vietnamese Automatic Speech Recognition.Google Scholar
- Awni Hannun, Carl Case, Jared Casper, Bryan Catanzaro, Greg Diamos, Erich Elsen, Ryan Prenger, Sanjeev Satheesh, Shubho Sengupta, Adam Coates, 2014. Deep speech: Scaling up end-to-end speech recognition. arXiv preprint arXiv:1412.5567 (2014).Google Scholar
- Geoffrey Hinton, Li Deng, Dong Yu, George E Dahl, and Abdel-rahman Mohamed. 2012. Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazine 29, 6 (2012), 82–97.Google Scholar
- Geoffrey Hinton, Li Deng, Dong Yu, George E Dahl, Abdel-rahman Mohamed, Navdeep Jaitly, Andrew Senior, Vincent Vanhoucke, Patrick Nguyen, Tara N Sainath, 2012. Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazine 29, 6 (2012), 82–97.Google Scholar
- Geoffrey Hinton, Li Deng, Dong Yu, George E Dahl, Abdel-rahman Mohamed, Navdeep Jaitly, Andrew Senior, Vincent Vanhoucke, Patrick Nguyen, Tara N Sainath, 2012. Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazine 29, 6 (2012), 82–97.Google Scholar
- Simon A Kingaby and Simon A Kingaby. 2022. Voice User Interfaces. Data-Driven Alexa Skills: Voice Access to Rich Data Sources for Enterprise Applications (2022), 3–14.Google ScholarCross Ref
- Larry R Medsker and LC Jain. 2001. Recurrent neural networks. Design and Applications 5, 64-67 (2001), 2.Google Scholar
- Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2023. Robust speech recognition via large-scale weak supervision. In International Conference on Machine Learning. PMLR, 28492–28518.Google Scholar
- Wojciech Samek, Grégoire Montavon, Sebastian Lapuschkin, Christopher J Anders, and Klaus-Robert Müller. 2021. Explaining deep neural networks and beyond: A review of methods and applications. Proc. IEEE 109, 3 (2021), 247–278.Google ScholarCross Ref
- Dang Dinh Son, Dang Xuan Vuong, Duong Quang Tien, Ta Bao Thang, 2022. ASR-VLSP 2021: Conformer with Gradient Mask and Stochastic Weight Averaging for Vietnamese Automatic Speech Recognition. VNU Journal of Science: Computer Science and Communication Engineering 38, 1 (2022).Google Scholar
- Pham Viet Thanh, Dao Dang Huy, Luu Duc Thanh, Nguyen Duc Tan, Dang Trung Duc Anh, Nguyen Thi Thu Trang, 2022. ASR-VLSP 2021: Semi-supervised Ensemble Model for Vietnamese Automatic Speech Recognition. VNU Journal of Science: Computer Science and Communication Engineering 38, 1 (2022).Google ScholarCross Ref
- NGUYEN Thi Thu Trang and NGUYEN Xuan Tung. 2019. Text-to-speech shared task in VLSP campaign 2019: evaluating Vietnamese speech synthesis on common datasets. Vietnamese Language Signal Processing. VLSP (2019).Google Scholar
Index Terms
- Vietnamese Voice2Text: A Web Application for Whisper Implementation in Vietnamese Automatic Speech Recognition Tasks: Vietnamese Voice2Text
Recommendations
HMM-Based Vietnamese Speech Synthesis
In this paper, improving naturalness HMM-based speech synthesis for Vietnamese language is described. By this synthesis method, trajectories of speech parameters are generated from the trained Hidden Markov models. A final speech waveform is synthesized ...
Modeling Vietnamese Speech Prosody: A Step-by-Step Approach Towards an Expressive Speech Synthesis System
Revised Selected Papers of the PAKDD 2015 Workshops on Trends and Applications in Knowledge Discovery and Data Mining - Volume 9441Attempts to add expressivity to synthesized speech is one of the main strategies in speech technologies. This paper summarizes our researches on modeling Vietnamese prosody, with the goal of improving naturalness of synthesized speech in Vietnamese, as ...
Vietnamese automatic speech recognition: the FLaVoR approach
ISCSLP'06: Proceedings of the 5th international conference on Chinese Spoken Language ProcessingAutomatic speech recognition for languages in Southeast Asia, including Chinese, Thai and Vietnamese, typically models both acoustics and languages at the syllable level. This paper presents a new approach for recognizing those languages by exploiting ...
Comments