AISpeech is the API behind a Chinese language and pronunciation learning website, AIChinese. You may know that Chinese is a tonal language. Where tones and inflection are used in English to convey emotion or differentiate between a declarative statement and a question, in tonal languages like Chinese, different tones on the same syllable can be altogether different words. The AISpeech API gives you access to a language processing engine which is capable of detecting spoken words and evaluating if the appropriate tone is used.
This API offering is a combination of true API and embeddable utility. This product is designed to take spoken word audio samples and return a score on the speaker’s pronunciation and tone. AISpeech provides flash objects for recording a voice sample and also for tone curve drawing. A tone curve demonstrates if the syllable was pronounced with an up, down, flat, or u-shaped inflection. The scoring part of this API is returned in JSON. Non-Chinese reading programmers will need to fire up Google Translate in order to get through the documentation. But we’re learning languages, and so we’re all about translation and breakin’ down that language barrier, right?
This API seems to be tailored more toward language learning applications. If I had to take a guess at their API strategy, I would say that AISpeech is hoping to become a point of reference for language learning sites.
AISpeech is one of seven speech recognition APIs, but the first to focus on learning a language.