Renewed sorting system using AI voice recognition at Ginza Cozy Corner distribution center

Advanced Media Co., Ltd. (Headquarters: Toshima-ku, Tokyo; Representative Director, Chairman and President: Kiyoyuki Suzuki; hereinafter referred to as "Advanced Media") is Ginza Cozy Corner Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; Representative Director and President: Norio Yamashita; hereinafter referred to as: We would like to report that we have renewed and introduced a sorting system that utilizes voice recognition at the Ginza Cozy Corner (Ginza Cozy Corner) distribution center, and that it has been highly effective.

Ginza Cozy Corner is a sweets shop that is celebrating its 400th anniversary this year, with approximately 70 stores nationwide, mainly in the Tokyo metropolitan area. More than 200 products are always on display at the store, including the popular jumbo cream puffs, a wide variety of fresh cakes, baked goods and chocolates that are perfect for gifts.

In 2010, a "sorting system" that utilizes voice recognition was adopted at the Ginza Cozy Corner distribution center with the aim of delivering products to stores more accurately and efficiently. By utilizing voice recognition, hands-free and eyes-free are realized, and compared to the touch panel format of the handheld terminal used previously, the work efficiency is increased by 1% and the error rate is reduced by 20% to 84 employees per hour. We were able to obtain a high introduction effect of 12 in 5.

In order to further improve business efficiency, we have been working with Ginza Cozy Corner's information systems department and logistics center to review specifications and have introduced a new sorting system equipped with the latest deep learning (DNN) compatible AI voice recognition engine on Android devices. did. The voice recognition development kit "AmiVoice SDK for Android" was adopted, and the Android application was redeveloped under the leadership of Hitachi Industrial Control Solutions Co., Ltd. Compared to before the renewal, work efficiency has increased by 20% and the error rate has decreased by 62% to 33 in 1 due to improved response speed due to improved device specifications and improved speech recognition accuracy due to the implementation of deep learning technology. did. Furthermore, the microphone that connects to the device used has been renewed to a wireless type, greatly reducing the risk of failure in the connection part that occurred when wired microphones were used.

Advanced Media will continue to actively develop new services that contribute to business improvement and efficiency through the use of voice recognition technology.

Above