AmiVoice API allows you to integrate high-quality speech recognition functionality into your services and products.
News
2026.02.09
[Feb 2 (Wed)] AmiVoice Developers Meetup: A seminar held exclusively here! The realities of implementing speech recognition from an engineer's perspective - The role of speech recognition in the age of generative AI and behind-the-scenes product development2026.02.05
[2026 Latest Edition] Comparison of speech recognition APIs from the seven major companies releasedA speech to text API that converts speech to text. By sending audio to the API endpoint, you receive a transcribed text of the speech. Speech-enabled applications such as for conference transcriptions and voice dialogue systems can be created.
In addition to the "general-purpose engine" that can be used in a variety of situations and businesses, we have engines that specialize in technical terms and industry terminology such as those in the medical field. By selecting the engine that best suits your usage scenario, the recognition rate can be significantly improved.
There are no initial costs and it is available on a pay-as-you-go basis. You can use high-quality speech recognition engines at low prices starting from 79.2 yen per hour. Each engine can be used free of charge for 60 minutes per month.
The Conversation_General-purpose engine is constantly updated to the latest version by the Auto-Training System (ATS). It can also recognize new words that have only recently appeared.
We handle everything in-house, from speech recognition engine development to service provision, and our technical staff will provide direct, free support for any technical inquiries, such as when introducing the API or when dealing with individual API-related problems after the start of operation.
Real-time text conversion (WebSocket/Synchronous HTTP)
Receive streamed speech recognition results in real time as the audio is processed.
Usage scenario examples
Batch Text Transformation (Asynchronous HTTP)
Batch convert huge audio files to text.
Usage scenario examples
You can register words in the user dictionary, such as product names and proper nouns, depending on the usage scenario. This improves the recognition accuracy of specific words and phrases, such as company terminology and names.
AmiVoice API automatically adds punctuation and question marks to the recognition results, helping you create more accurate and understandable spoken sentences.
Fillers such as "ah," "urm," and "hmm" are automatically deleted (you can also set them not to be deleted). This reduces the effort required to correct recognition results.
This function identifies who spoke and when for audio containing multiple speakers. It identifies and labels speakers without prior training.
You can use it with confidence because inappropriate terms (sexual language, discriminatory language, etc.) in official situations such as business settings will not be converted into text.
The speech recognition results are given pronunciation that is estimated from the speech. This can be used not only for display in apps for children, but also for data matching in subsequent systems such as CRM and product databases.
The system analyzes the speaker's emotions from the voice and outputs 20 parameters, such as joy, anger, stress, dissatisfaction, and expectation, approximately once every two seconds. It allows for more detailed voice analysis, such as being able to grasp the emotional ups and downs hidden behind the expression "thank you" for example. Sentiment analysis results can be used in a wide range of situations, including contact centers, marketing, human resources, and medicine. We provide ESAS technology. Based on the latest sentiment analysis engine provided by Israel's Nemesysco, it was developed by ES Japan through verification and research of 3 million audio sources in Japan. You can use it with the asynchronous HTTP speech recognition API. You can use it for free for up to 60 minutes per month, so please feel free to try it out.
60 minutes free per month on all engines
Start Using APIPlease check out our webinars to learn more about our services and speech recognition technology.
It achieves over 80% accuracy in name recognition using kana‑based input and meets the required security standards.
Conversations with patients are converted into text and linked to a generation AI. Highly acclaimed for its high recognition accuracy of medical terminology.
We have developed a product that enables accurate and fast subtitling of live broadcasts without the need for specialized skills.
