What projects allow you to recognize the speaker by voice?

Question

Google

Actually, the question is in the title ...
I'm doing a small homemade product for a "smart home" and I want some kind of API to be able to distinguish between family members.
I would like to check the rights to any actions (for example, only I can turn off the light in my room).
The API must return the speaker's internal identifier.
I want to tie it to api.ai :)
The answer was found in the SpeechKit Cloud documentation:
https://tech.yandex.ru/speechkit/cloud/doc/guide/c...

Answer 1

I will advise a slightly different approach: this is the intersection of the corridors of the average signal carrier for peak values ​​​​(or the corridor of deviations).
1. Take the normalized signal (x-axis: time, y-axis: amplitude) and plot the average curve.
2. Then, we make an allowance for errors: we increase the thickness of the curve and get a confidence zone in which we will consider the signal.
3. Then, we cross the zone from the just received input signal (from the microphone) with all stored reference zones (one such zone is one pre-recorded human voice with the applied transformations from paragraphs 1 and 2).
The maximum coincidence of zones when crossing (overlapping) will give the desired family member.

Answer 2

https://habrahabr.ru/post/144491 - read superficially for a start))