AI, Inc. is a Japanese company specialized in providing and developing speech synthesizers, having its own product called AITalk, the system upon which VOICEROID –as well as other products like Gynoid Talk– are based. It is located in Tokyo, Japan, and is the main creator behind the Kotonoha sisters and Iori Yuzuru.
History[]
AI, Inc. was founded on April 1, 2003 with the goal of providing and developing speech-synthesis software with their respective voicebanks, in addition to consulting businesses about such topic: speech synthesis. Its goal was to provide the best speech-synthesis software around the world and create a culture around sound information so that it can contribute to our daily lives around development of speech-synthesis software and services.
In late 2019, the company announced that it would be upgrading AITalk's engine from AITalk®4 to AITalk®5.[1] Unlike previous iterations of the engine which used corpus-based speech synthesis and phoneme and rhythm dictionaries, AITalk®5 would use deep neural network speech synthesis instead. This would allow AITalk®5 vocals to speak more naturally and express multiple emotions more smoothly. Without a dictionary that a corpus-based speech synthesis would require, the deep learning model would require less recordings and thus be less expensive for customers as well.[2]
On June 26, 2020, Kotonoha Akane & Aoi were announced to receive a singing library for Synthesizer V Studio, making them the first entry as Synthesizer Vs from AI, Inc. as well as the company's first singing voice library overall.[3]. On July 30, the pair released as a single shared Standard voice database where producers adjusted the voice parameters to represent the sister they wanted.
Continuing development of speech synthesis, in September, it was revealed that the commercial brand of AITalk®5 would be called A.I.VOICE.[4] A.I.VOICE was scheduled for a February 2021 launch beginning with the Kotonoha sisters and Iori Yuzuru, with more characters to be announced later for July and December 2021. In December 2020, details about the Kotonohas' and Yuzuru's A.I.VOICE banks were revealed, along with a release date of February 22, 2021.[5]
A.I.VOICE is planned to have English and Chinese voices by December 2021, along with singing synthesizer voicebanks.[6]
Voice databases[]
Web Synthesizer V[]
Voice databases developed by AH-Software Co. Ltd. in collaboration with AI, Inc.
Synthesizer V Studio[]
Voice databases developed by AH-Software Co. Ltd. in collaboration with AI, Inc.
References[]
- ↑ https://ssl4.eir-parts.net/doc/4388/tdnet/1751595/00.pdf
- ↑ https://vocaverse.network/threads/aitalk-getting-deep-learning-based-engine-update-april-2020-a-i-voice-brand.2239/#post-15621
- ↑ https://twitter.com/ahsoft/status/1276352204547358720?s=20
- ↑ https://www.ai-j.jp/20200930_2
- ↑ https://www.ai-j.jp/20201218
- ↑ https://www.ai-j.jp/20201109