Raven without Rhasspy

[romkabouter]

@principemestizo

Don’t you think the process would be faster without being recorded? That is, that he understands all the words you say and that he immediately recognizes the sentence.

Highly unrecommended. Number 1 reason: Rhasspy might act on something you do not intent to be acted upon.
That will lead to very annoying situations, which you can probably think of by yourself

[principemestizo]

I think you misunderstood what I said, what I propose is the following:

1. Recognize the wakeword
2. The SST is activated. Let him recognize all the words he says below and put them in a list (split). With the condition if 2 or 3 words of a sentence are said, it knows that it belongs to X sentence and acts immediately without having to record and recognize.

Bad idea?

[romkabouter]

Ah ok, sorry about that
That might be harder to do, but I do not know that

[principemestizo]

@synesthesiam

Also, there really is no limit at 3 WAV files. It’s the minimum recommended, but you can have as many examples as you want. Without template averaging, this will be slower. With it, there should be no performance difference (though it may be less accurate).

The command that allows to use multiple keywords, may I know how it works? I mean, is it like I have multiple wavs in a single folder or is it different at the code level?

[synesthesiam]

@principemestizo

I mean, is it like I have multiple wavs in a single folder or is it different at the code level?

Each keyword needs to be in a separate folder, but you can have as many WAV files in there as you’d like. So keyword1 could have 10 WAV files, keyword2 only 3 WAV files, etc.

If you do --average-templates each folder of WAV files will be averaged and Raven will run faster. Without --average-templates, each WAV file is checked against the current audio in a loop, so this will slow down as you have more files.