ELSA, a mobile app that uses Artificial Intelligence and speech recognition technology to help language learners improve their English pronunciation, today announced that it has raised US$3.2 million in pre-A round of financing, led by Southeast Asian VC firm Monk’s Hill Ventures.

As part of the agreement, Monk’s Hill’s Managing Partner Peng T Ong, will join ELSA’s board.

ELSA, which stands for English Language Speech Assistant, will use the capital to hire top-tier talent to develop new AI capabilities, open its voice recognition API, and expand its speech recognition technology to teach new languages beyond American English.

Also Read: Meet ADSvokat, the Indonesian startup that will help you make money from the sticker on your laptop

Additionally, the startup will create a platform to support English teachers and schools around the world to integrate ELSA into their curriculum and help students with their speaking skills.

Using ELSA’s voice recognition technology, language learners can improve their accent, pitch and intonation when speaking English. The app listens to the way language learners pronounce words, sentences or conversations to pinpoint exact errors and provide real-time, accurate feedback on their pronunciation mistakes with specific suggestions on how to improve.

For example, ELSA teaches English learners how to move their tongue or their lips differently in order to fix pronunciation errors.

“We’ve built a very large speech corpus of non-native speakers from around the world learning English. With this very valuable data set that we’ve collected and trained over the past few years, we continuously improve our advanced pronunciation assessment engine using state-of-the-art speech recognition and deep learning technologies. The support of Monks Hill Ventures is crucial as we execute our plans for growth and development to new languages, educational platforms and extensive AI capabilities,” said Vu Van, Co-founder and CEO of ELSA.

ELSA uses recurrent neural networks (specifically LSTM networks) to model American speech and detect user’s pronunciation mistakes with 90 per cent accuracy, it claims.

ELSA has earlier raised funding from investors including 500 Startups.

Correction: A previous version of this article claimed ELSA had 1.5 billion users. That number was incorrect.