TECH Meets BUSINESS

In brief, we have released an AI platform devoted to voice recognition, that commercializes the most advanced technologies at the research level, obtained by technology transfer from AIST.
In many businesses various conversations and meetings, etc., are held every day either by telephone or face-to-face. If the non-structured data of voice can be converted into structured data, in other words converted into text using voice recognition, then it is likely that information that is disintegrating or buried within the company can be deployed for productive activities in the company.
In recent years there has been significant progress in artificial intelligence (AI) and deep learning. Image recognition technology is being applied to image diagnosis, automatic driving, etc., and research is in progress in Silicon Valley on automatic driving level 4. However there is still scope for improvement in the recognition accuracy of voice recognition technology. At present voice assistant and communication robots such as Siri from Apple, OK google, etc., have started to be used by many people. However this is ok for private use, but its use in business where high accuracy with non-steady noise in the background and security are required is still some way off. In particular voice recognition in the Japanese language is very difficult. People do not talk grammatically, myself included (laughs), and there are dialects. Conversely it is a field with high technical potential, so we are working to create a new market as fast as possible, with new and the latest voice recognition technology and by adapting various other technologies.

At present we broadly provide 3 types of service and solution. The first is “VCRM”. This is a solution for financial institutions for customer relationship management, and at present we provide services to the Corporate Sales Departments of MUFG Bank, Ltd. This is to increase the efficiency of work at the bank, but it was also introduced with the idea of improving customer service by solving issues and making proposals regarding sales and operations, using customers’ voice that has been converted to text.
Next there is “VContact”, a solution for call centers. Every day call centers have a massive volume of telephone calls with customers, and by introducing this solution the conversations are converted into text in real time, and in addition a summary of the contents of the call can be produced automatically. This is of value not only for improving efficiency, labor-saving, and prompt and appropriate dealing with complaints, but also business issues can be extracted from the collected voices of the customers, and utilized for sales, product and service development, as well as marketing activities. This solution can be based on our own server (on premises), or anywhere in the cloud, so we can provide the service in accordance with the customer’s situation and needs, required security level, etc.
The third is “VRobot”, a solution provided to many robot manufacturers and device manufacturers as a mechanical listening sense for robots and IoT devices. In addition there is the “VCrowd” service that produces text at low cost and high quality in a short period of time by voice recognition + human power, “VBox” automatic voice recognition from video and voice files, etc. We provide services that allow our latest technologies to be freely tested, and through these services our voice recognition engine is strengthened using actual voice data every day.

“The Voice JP” incorporates the minimum necessary collected volume, high processing speed, analysis, and utilization for analysis of voice big data. The accuracy of the initial collection greatly affects the quality of the subsequent processing, so the highest advantage is the “accuracy of the collection”, in other words accuracy of the voice recognition, and it must be a massive strength. In the “VCRM” mentioned previously, we can boast of 90-95% input accuracy.
Furthermore, after introduction of the service, the system continues to automatically learn unknown and new words. Day-by-day it continues to evolve learning from company data, such as particular terminology and phrases used in the company, the names of new products and services, new employee names, etc., and learning from text on the internet, as well as field-specific terminology.

Let us consider the example of call center solutions, which is easy to understand. In the very near future it will be possible that about 30-50% of inquiries can be resolved by automatic dialog without the intervention of an operator. Also, as I touched on previously, the call center will become the base for the frontline of the marketing function of companies, and companies will create new products using VContact from dialogs with customers. A world will be created where planning and sales strategies for new products are produced extremely efficiently. Our products must also evolve further in accordance with the dramatic changes in public activities. At present the use of monitoring (crime prevention) cameras is spreading, but we are seriously considering the future in which conversations will be monitored, and the optimum next action will be determined.

I originally worked in the software company Fujisoft Incorporated. I considered that a venture company where decision making could be made nimbly and speedily was more appropriate for the challenges of breakthrough fields such as voice recognition technology, and my apologies to Chairman Nozawa who had mentored me until then. I had interactions with AIST from my days with Fujisoft, and for a long time I was interested in voice recognition technology. This led to me starting the company, without too much agonizing. I constantly think that it would be nice if in the future the company is able to report to Chairman Nozawa.

In the future AI including voice recognition technology will greatly change the form of business. I think the most important thing is to refine the advanced technologies of AIST and the existing technologies in the next 5 to 10 years, so that replacement of the existing traditional market is brought about at least in part by our work. That is the strength of a venture company, and I see it as our mission.

* The contents of this article are based on information as of January 13, 2017.