Voice Recognition System "TeamLog" Born from the Wishes of the Hearing Impaired Officially Launches on April 13th! Supports Transcription of Multi-Person Conversations and Noisy Environments, and Real-time Translation of Multiple Languages.

Hearing Lab Co., Ltd. (Head Office: 1-5 Toei-cho, Gifu City, Gifu Prefecture; Representative Director: Hiroko Shimizu) will release the new real-time voice recognition system "TeamLog" on April 13, 2026. This system uses a unique mechanism with multiple microphones to support simultaneous transcription for up to 9 people, transcription accuracy that does not decrease even in an 80dB noisy environment, and simultaneous translation of conversations involving multiple languages.

Currently, efforts are underway to provide more reliable information support for the hearing impaired by combining PC captioning, and to support the education of foreign children who have not yet learned Japanese by improving translation accuracy.

■ First Voice Recognition Research Triggered by Corona

Hearing Lab Co., Ltd. was established in 1993, with "Gifu Prefecture Association of Parents of Children with Hearing Impairments" as its predecessor. Our roots lie in the efforts of parents of children with hearing impairments to research the development, hearing, and hearing aids for their own children.

We began our voice recognition research during the COVID-19 pandemic. While many people struggled with the pandemic, the hearing impaired fell into severe communication difficulties as everyone started wearing masks. For the hearing impaired, who rely on lip movements as a visual aid for hearing, having their mouths hidden by masks was a critical problem.

"I can't hear conversations at all," "Even shopping is a struggle," "I can't ask people to take off their masks, what should I do?"

Many voices reached us, and we started researching voice recognition that could convert speech into text, wondering if there was anything we could do.

■ Serious Challenges of Previous Voice Recognition

Although we started voice recognition research with this background, the more we worked on it, the more its difficulties became apparent. Misrecognition is common in voice recognition, but since the hearing impaired cannot confirm misrecognitions with their ears, conversations become incomprehensible if there is even a slight error.

Furthermore, most transcription tools perform transcription using the built-in microphone of smartphones, etc. Therefore, in noisy environments or situations where many people are speaking, which are often problematic for the hearing impaired, voice recognition frequently does not function properly due to the influence of other sounds.

■ Proprietary Mechanism Under Patent Application

TeamLog solves previous challenges with a proprietary structure using multiple directional microphones.

I. Simultaneous Transcription for up to 9 People

Supports simultaneous transcription for up to 9 people by using multiple microphones. It can transcribe all utterances even when speech overlaps or interruptions occur. By assigning names to microphones, we have created a design where "who said what" is clear at a glance.

II. High Recognition Rate and Support for Technical Terms

High accuracy is achieved by adopting a high-precision voice recognition engine. Furthermore, by creating a word dictionary, it supports technical terms and personal names. It also includes dedicated engines for fields like medicine and finance, enabling high-precision transcription even in specialized meetings.

III. No Accuracy Reduction Even in Noisy Environments Exceeding 80dB

By adopting microphones with strong directivity, we have succeeded in minimizing the influence of noise and other people's voices. Transcription accuracy is virtually unaffected even in noisy construction sites, or in places like cafes and izakayas where people are talking loudly nearby.

■ Currently Focusing on Support for Foreign Children

We are currently focusing on supporting the education of the rapidly increasing number of "foreign children who have not yet learned Japanese" in Japan. Existing transcription tools have low recognition accuracy in classrooms where various voices are flying around, but TeamLog can achieve high accuracy.

Leveraging this high recognition rate, we are continuing development and verification to enable real-time translation into various languages.

■ Also for Meeting Minutes and Simultaneous Interpretation

Although TeamLog was originally developed for information support for the hearing impaired, it also has functions as a convenient tool for meeting minutes and simultaneous translation for a wider audience. This is because if it were only useful for the hearing impaired, it might be difficult to get it adopted – it would just be something "hearing impaired people asking for it to be introduced."

Nowadays, with the mandatory reasonable accommodation, it's easier for companies to accept requests from people with disabilities. However, voice recognition often has high hurdles in terms of cost and security, making it difficult to implement solely for disability support.

Therefore, we have incorporated various features to make the service effective for companies and organizations by balancing useful functions with high security.

Feature List

◇ All recognition results are saved locally for security

◇ Output in CSV and PDF, usable as meeting minutes

◇ Offline recognition function (planned) that does not record or communicate

◇ Simultaneous interpretation function to recognize and translate multiple languages simultaneously

◇ Real-time correction to achieve 100% transcription accuracy

◇ Real-time display on other terminals via sharing function

◇ Translate to preferred language on shared terminal

We aimed for a system that more people can use through these features.

■ Pricing

The pricing for the TeamLog voice recognition system is as follows:

〇 Corporate Plan

Initial Cost: None

Basic Fee: ¥10,000/month + voice recognition engine usage fee (usage-based)

〇 Individual Plan

Initial Cost: None

Basic Fee: ¥1,000/month + voice recognition engine usage fee (usage-based)

〇 Support Plan

For individuals with disability certificates for hearing impairment and organizations supporting the hearing impaired.

Initial Cost: None

Discount on individual/corporate plan fees.

*Currently only iOS version available (Android and PC versions under development)*

*Usage fee for voice recognition engine is approximately ¥100 for 3 hours on average.*

*Separate purchase of microphones etc. required (general-purpose items can be used)*

■ Future Outlook

TeamLog has received significant response even before its release, and trial tests are currently being conducted at employment facilities and schools for the hearing impaired. In recent years, we have been working on support for foreign children who have not yet learned Japanese, and on building a real-time correction system using the included editing software. We will continue to aim for further accuracy improvement, functional enhancement, and adaptation to a wider range of fields.

■ Company Overview

Company Name: Hearing Lab Co., Ltd.

Address: 1-5 Toei-cho, Gifu City, Gifu Prefecture

Representative: Hiroko Shimizu

Business Activities: Sales of welfare equipment, system development, information support

Contact: teamlog@hearing-lab.jp

Official Website: https://hearing-lab.jp

Product Page: https://teamlog.hearing-lab.jp