With the continuous improvement of speech recognition technology andcomputer vision technology, modern medicine is entering a comprehensive digital era. The pursuit of high-quality training data is the creed of the artificial intelligence industry, and obtaining a more accurate speech machine model is inseparable from the continuous supply of speech data. This article introduces:
- What is speech recognition technology
- How speech recognition technology is used in the medical industry
What is speech recognition technology
Simply put, speech recognition technology is the process of converting speech into text. In recent years, speech recognition technology in the medical field has significantly improved its ability to understand medical terminology and regional accents and dialects.
How speech recognition technology is used in the medical industry
A few years ago, Peking Union Medical College Hospital conducted a survey on document entry workload. The results show that more than 40% of doctors in my country spend about 4 hours a day in front of the computer typing text, and more than half of the doctors spend 40% of their total working time doing text typing every day. Doctors generally have a strong need to improve the efficiency of medical record entry. Speech recognition technology is a relatively mature solution. In many European and American countries, speech recognition technology has been used in the medical field for more than 10 years. In recent years, the proportion of voice recognition input applications in clinical practice in the United States has reached nearly 20%.
MediaInterface is one of the representative speech recognition technology companies. Centered in Germany, MediaInterface has provided voice technology solutions to medical institutions in many European countries for more than 20 years. As its name suggests, its core product Speaking is an efficient tool that uses speech recognition AI technology to support medical documentation. Currently, this product has helped 600+ hospitals, 700+ medical institutions, and 75,000+ users achieve faster and higher-quality medical document workflows. With more than 20 years of successful experience in many European countries, MediaInterface hopes to expand its business to France. However, in the process of business expansion, they encountered the problem of insufficient French data resources.
Despite more than 20 years of industry experience, MediaInterface does not have much data in French. To expand into the French market, MediaInterface must establish a comprehensive French vocabulary foundation. They urgently need professional French "dictionaries" with high-quality phonetic transcription capabilities. The largest data gaps in this "lexicon" are the French names and places frequently cited in patient health information. But this data is difficult to access: Europe’s General Data Protection Regulation stipulates that health data that can be collected must be anonymized, so it does not include names of people and places. MediaInterface must seek professional external resources to help them fill these important data gaps while complying with local data regulations.
We were expanding into new markets. Although we have advanced technology and software, the lack of data resources makes it difficult to implement our products locally. Appen's French data resources helped us promptly. ——Ines Wendler MediaInterface Product Manager
MediaInterface and Appen got connected at the 2015 INTERSPEECH Voice Technology Conference. The expansion of the French market has contributed to the cooperation between the two parties. Currently, MediaInterface has obtained approximately 21,000 French names and 14,000 French place names through Appen, thus filling the most critical data gaps. Appen helps MediaInterface develop products for a certain concentrated demand through a rich variety of finished product data sets, thereby expanding into new markets and enhancing its ability to enter new markets in the future. Multiple market possibilities. Currently, MediaInterface has fully covered the basic spoken speech recognition needs of medical institutions in France, providing French customers with high credibility. With rich data resources, the SmartLearning function in SpeaKING also allows users to add their own text to personalize existing vocabulary, thereby improving the effect of speech recognition by adding data to the basic AI model. For MediaInterface, Appen Dataset has equipped it with tools to expand its customer base while also improving data quality and customer experience. Not just MediaInterface, Appen's finished product data sets have helped many companies achieve rapid deployment with higher product accuracy.