AmiVoice SDK

AmiVoice SDK, for which over 5,000 licenses have been installed, is used for voice input and speech-to-text conversion in various usage scenarios due to its high recognition accuracy. The library also provides audio recording functions, making it easy to develop native apps for various operating systems. By linking with the speech recognition server (AmiVoice API Private), operation is also possible regardless of device specifications.

Service Intro Webinar

Get Materials

AmiVoice SDK Features

Available offline
More secure operation
In the standalone setup, speech recognition can be used even in environments where it is difficult to connect to the Internet from a security perspective, such as in medical settings.
Examples of use in the medical field
Word and command recognition also possible
For voice input and voice control
It is also ideal for recognizing combinations of product names and numbers, numerical values, letters, and commands such as "confirm" and "next" (rule grammar recognition). Naturally, it also recognizes sentences with high accuracy.
Noise-resistant engine
For developing outdoor/factory applications
Highly customizable, allowing adjustment of the recognition parameters to suit the site, and recognition of technical terms by pre-registering them. Can also be used for embedding in devices used in manufacturing sites and outdoor inspection and maintenance work, and in app development.
SDK case studies
Wide-ranging support options for development with peace of mind
For technical inquiries about how to use the SDK, ①technical staff will directly support you ②in Japanese. We also provide training on how to use the SDK upon request.
About the support menu (materials download)

Choose the speech recognition type that suits your needs

Type 1: Rule grammar (word/command) recognition
It only recognizes standard phrases and words that have been set up by you in advance. Use it when you want to operate the device by voice or input pre-prepared words and phrases by voice.
About the rule grammar engine
Type 2: Dictation (text) recognition
It converts speech, such as "spoken language" or "written language," that does not fall within the scope of the rules into text. It displays the text that is calculated to be closest to the spoken content.
* Standalone recognition speed depends on device specifications. * If you prioritize recognition speed, we recommend using AmiVoice API Private.

Click here for a list of engines

Customizable engine

In addition to the "general-purpose engine" that can be used in a variety of situations and businesses, we have engines specialized in technical terms and industry terminology, such as those in theMedical field. We can also customize existing engines for our customers to achieve maximum rates of recognition.
About customization
Speaker diarization function

This function identifies who spoke and when for audio containing multiple speakers. It identifies and labels speakers without prior training.