Real-time audio (voice) processing, Arduino or raspberry?

Question

Arduino

Good day, you need to implement real-time sound processing, in this case, the circuit should receive a sound signal via wifi (or from the local host of my computer - both options are desirable), and convert it to text, and also do the reverse of this operation. The question concerns only the technical part, as far as I know arduino is too stupid for such operations, but perhaps with some different modules, you can make it do it. I had no experience with such things, can anyone tell me what can be used for these purposes?

Answer 1

voice recognition on arduino unlikely.
Yes, and raspberries - I strongly doubt it. Another thing is that on raspberries you can send this stream to the cloud and get the text already.

Answer 2

Arduino or raspberry?

Raspberry of course.
And the question is not only performance, but the fact that with raspberry you can use the whole set (to choose from) of all open projects for Linux.

Answer 3

The task of converting voice to text consists of three stages. First, phonemes are recognized, then phonemes are extracted from the stream, then sentences are formed from words. The experience of using Android and Siri shows that even with three stages, it is far from always possible to turn a voice message into a text message without errors.
So take an Arduino and teach it to recognize phonemes. It will be very cool, but don't tell anyone that you originally wanted to convert voice to text.