Creating a wikipedia skill : match any regexp?

[snoopfab]

I’m trying to build a wikipedia skill for my rhasspy assistant.
I would like to say something like “wikipedia $(some long search query)” but i don’t know how to match the long search query which can be anything.

[Wikipedia]
Wikipedia [.*] {searchQuery} (where .* can be any python/javascript regex)

I know i can switch rhasspy to an open transcription mode but wouldn’t i loose all my other intents commands ?

Is there a simple way to do something llike this ?

By the way, thanks again for this huge and highly qualitative job that is rhasspy !

[Enc3ph4l0n]

It depends on your definition of simple

If you haven’t seen it already I’d encourage you to look at this thread:

Recognized untrain sentences / words

Hi, Is there a way to get a slot for an unknown part of a sentence ? My idea would be to do a reminder intent. I can define the start of the sentence, but would never know what I will say to remind. example : Hey Rhasspy, remind me to call my wife in two hours Hey Rhasspy, call me back in 30 minutes to shutdown the oven Hey Rhasspy, bring me back to take my son out of freezer in ten minutes etc … Any idea ?

There are a few “solutions” discussed, ranging from using Google TTS to process part of the audio, to having a second environment set up as open transcription so you have the best of both worlds.

[kart-able]

It seems you found your “simple” way, but I have not. Can you explain?

[snoopfab]

As Enc3ph4l0n said there is no “easy” solution to this one. However in the mentioned post there is one suggestion that I intend to try : use a second installation of rhasspy in open translation mode with a specific wakeword (wikipedia in my case) to handle this intent. I setted the solution flag as my question was answered even if the answer was negative. For the moment and as a workaround i use pregenerated slots for artists / cities / painters … that I want to query in wikipedia.

[kart-able]

Great !
I would definitely be interested in a Wikipedia skill. But without the creation of slots beforehand.

I did not fully understand what would be required: a second installation of Rhasspy???

[snoopfab]

i’m afraid that the route of open transcription mode may imply a second installation of rhasspy.

As described in the rhasspy documentation :

“If you just want to use Rhasspy for general speech to text, you can set speech_to_text.kaldi.open_transcription to true in your profile.” (Given that you use kaldi or pocket sphinx for tts)

However, i found another option in the rhasspy doc that may be worth investigating further :

“A middle ground between open transcription and custom voice commands is language model mixing . During training, Rhasspy can mix a (large) pre-built language model with the custom-generated one. You specify a mixture weight (0-1), which controls how much of an influence the large language model has; a mixture weight of 0 makes Rhasspy sensitive only to your voice commands, which is the default.”