Photo: Daisuke Yoshida, Representative Director of AI, Inc.
We are living in a society surrounded by human-like voice spoken by robots, such as speeches announced in train stations/buses and of vending machines, car navigation systems, game devices, smart phone applications, automated phone answering, etc. We interviewed President Daisuke Yoshida of AI, Inc., which has powerful technologies and is enhancing its potentials in speech synthesis system developments.
Photo: Entrance of AI, Inc. head office surrounded by clean atmosphere. Yoshida says it offers comfortable development circumstances mostly free from other firms’ constraints as it focuses on propriet
President Yoshida says he first felt great potentials of speech synthesis technologies in around 1999 when he worked with Advanced Telecommunications Research Institute International (ATR/Kyoto Prefecture). It is an international R&D base pursuing leading-edge researches in telecommunication areas through collaboration among industries, academia and public sectors.
“I was so captured by the potentials of converting texts to voice and synthesizing any individual’s voice, and I determined to diffuse the technologies to the general public. The first step I took was weather forecasts by synthesized voice proposed to Hankyu Corporation when I knew it was renewing its websites. This was in fact materialized after the firm welcomed my technical idea that synthesized voice of Takarasiennes could present weather forecasts. This proved very popular among viewers and was highly appreciated as the firm’s websites at that time consisted of only still pictures and texts with no videos contained.”
*Note: Stars of Takarazuka Revue, Japan’s most famous female-only entertainment group
This success triggered President Yoshida to move from Kansai to Tokyo and after some meanderings he spanned off from ATR to start AI, Inc. in 2003. Since then, it has been enhancing its presence as a unique company specialized in speech synthesis in Japan.
Photo: Popular AITalk® Series have been employed by more than 500 customer firms.
Our representative product is the AITalk® Series born in 2006. The feature is an implementation of “Corpus-Based Voice Aynthesis Method,” a synthesis technology using actual human voices, which produces natural voices very apart from conventional machinery automated voices. You can experience this technology with demonstrations on our websites, which we believe many should be impressed.
Its mechanism is as follows: It creates a voice dictionary from recorded voices reading a specific script which contains a lot of vowels and consonants (such a sentence as “All the realities were skewed toward one’s desirable direction.”). Secondly, the system analyzes the text and picks up the most appropriate voice from the dictionary judging from tones and intonations. Finally, it outputs series of voices synthesized with sound waves adjusted by signal processing technologies.
Photo: A row of award certificates at the entrance
Among recent products, what attracted people’s attention was our AITalk® applied to voice of the Matsuko-Roid, a robot of popular talent Matsuko Deluxe. It uses her recorded voices actually reading out texts. Also widespread are our products in a variety of fields including voice of Softbank’s amenity robot Pepper, medias such as Fuji TV news reading, announcements in stations and airports and e-learning materials. Our product lineups are rich, currently consisting of 15 and more languages and 40 and more speakers.
“We are admittedly not the only one with speech synthesis technologies, but AITalk® is applied at such very attractive costs as about \400,000 at the lowest. Speakers’ workload is also light as the time they need for reading out texts is short. Our peers would charge 10 times more and need longer time. We believe these strengths have been proving superiority of AI, Inc.”
What President Yoshida says should be confirmed by many kinds of social appraisals AI, Inc. has gained. In these days the Company has been honored by receiving many awards, including the Japan Venture Award 2014, the Officer-General of the Small and Medium Enterprises Agency Award, the Technology Development Award from Acoustical Society of Japan, the A-Rank Certification by Kyoto Ventures Mekiki (discerning) Committee, the Finalist of Entrepreneur Awards Program “EOY 2014 Japan” and the Grand Prize of Tokyo Venture Technology Award. Also the Company’s product was selected as a certified product under the Certification Program for Trial Ordering by Tokyo Metropolitan Government.
Photo: Added to AITalk® Series, VOICEROID Series offer selections of character voice originally developed by AI, Inc.
“Our locating in Tokyo offers not a few advantages, but one of the most significant is existence of a lot of development firms and end-users. More than anything, the City is full of manufacturing spirits of many firms aiming to challenge something new, which is the very environment we appreciate much.
Also very helpful was a quick reaction of Credit Guarantee Corporation of Tokyo upon our application for its loan programs.”
“There are a lot more things we can do with speech synthesis technologies in such favorable environment,” President Yoshida asserts.
“Such examples include making an archive of synthesized a child’s voices at younger age as a growth record and mails received by grandparents in a remote place are read out with synthesized grandchild’s voice. To cultivating 21st century cultures using AITalk® as a core—this is what we hope to realize.”