"Based on patent filings, Q.ai has built machine-learning tech for audio and “silent” voice input, including systems designed to understand to improve communication in noisy or difficult environments."

Good? Based on my recent experience the native iOS functionality around voice activity detection, echo cancellation and transcription is horrendous. If they could ship devices that had anywhere near the quality you can get with a bespoke stack to handle the audio data it would be impressive.

dupe: https://news.ycombinator.com/item?id=46814773