Automatic Speech Recognition

The ASR that understands the Taiwanese accent, helping you to convert multimedia and conversations to text.

Connect with a question-answering engine to provide voice interaction in the applications. Help enterprises to achieve digital transformation.
Record meetings and connect with other NLP technologies to organize discussions into easy-to-use information and knowledge.
Convert multimedias to text, so that you can understand the events from different sources and facets.

Even code switching speech can be dictated easily

Have immediate support from Taiwan’s team. Both Mandarin-English and Mandarin-Taiwanese speech can be accurately transcribed. Compatible with various scenarios with customized language or audio models to improve accuracy.

Yating’s speech recognizer has excellent anti-noise robustness

Our ASR model is developed for real-world scenarios. That means we use dataset from real use cases, select the most suitable deep learning archetetures for the scenarios and deliver industrial leading accuracy.

Easily transcribe audio and video files

Support a variety of formats. Include MP3, AAC, WAV, FLAC, YouTube, etc.

立即試用

API 用途

語音轉文字

我們與音樂之間的互動，不該只有被動的「聆聽」！當人工智慧精準掌握情感與溝通，即時創作獨一無二且有效的聲響設計，音樂的起伏走向由你主動掌控，人人都能有效地透過音樂表達自我，甚至成為個人化音樂創作，讓音樂即是生活！我們與音樂之間的互動

語音轉文字

產品價格

透過採用雅婷 AI 技術的 API 將語音準確轉換為文字。透過採用雅婷 AI 技術的 API 將語音。透過採用雅婷 AI 技術的 API 將語音準確轉換為文字。透過採用雅婷 AI 技術的 API 將語音。

查看詳細價格

You can either upload files or live streaming
Our ASR can dictate live streaming audio from the microphone or recorded audios.
Support multilingual and mixed language recognition
Mixed languages is pretty common nowadays. We provide Mandarin-English and Mandarin-Taiwanese code-switching recognition. English and Japanese are also available.
Build and support locally
The dataset is collected from real scenarios. Hence, our ASR’s anti-noise robustness is quite nice and can identify various accents and slang.
Language model customization
Provide meeting related data to improve the recognition accuracy. The data can be historical meeting records, meeting minutes, agendas, and participants names.
Customized Dictionary
Custom vocabularies such as names and industry specific terms, to increase the opportunities to recognize these words correctly.

Testimonial

National Theater & Concert Hall/ National Kaohsiung Center for the Arts

Two major art and cultural venues in Taiwan have adapted Yating’s speech recognition service in the performance narration and other events. Our ability to teach machines the terms and lingos of the art industry allows us deliver highly accurate code-switching real-time captions.

Taiwan Taipei District Prosecutors Office

Taiwan Taipei District Prosecutors Office uses Yating’s speech recognition technology to convert conversations among the prosecutor, the attorney and the defendant to text real-time, easing the court clerk’s burden. Considering investigation records are not open to the public, the system was deployed on-premis, hence, extremely confidential.

National Communications Commission

Our speech recognition model was developed to understand Taiwanese accent. It also offers excellent anti-noise robustness. Therefore, NCC uses our solution to convert news reports or vox pop to text for media monitoring. Through scientific analysis, NCC issues a monthly report online, urging audiences to supervising medias.

The Ministry of Justice

The Ministry of Justice has set up a speech recognition system, which is powered by Yating, in the prosecutors office. By using it, various teams generate meeting transcriptions effectively. The prosecutors can also inquiry audio evidences efficiently by referencing corrosponding transcriptions, instead of listening to the audio entirely.