Speech-to-phrase not acting on intent

Configuration
1

I added speech-to-phrase and configured it. However, it seems to be unable to act. On giving a command, I get a red led back on the voice assistant preview. AI suggested installing Rhaspy, so I did that and there was not much to configure. However, I rebooted HA and still the same result. I’ve also watched some set up videos and they’re not doing anything that I haven’t already done, but I must be missing something obvious somewhere.
The log appears to know that my intent is to change the office light to 50, as the spoken phrase works fine when typed into the assistant and also when I use faster-whisper … just not speech-to-phrase.
Any ideas on what I’m missing please?

DEBUG:root:Selected model by language: en_US-rhasspy
DEBUG:root:Using cached transcriber
DEBUG:speech_to_phrase.transcribe_kaldi:Stream ended
DEBUG:speech_to_phrase.speech_tools:lattice-to-nbest --n=3 --acoustic-scale=0.9 ark:/tmp/tmpv40kafsb ark:- | nbest-to-linear ark:- ark:/dev/null ark,t:-
DEBUG:speech_to_phrase.speech_tools:/usr/src/tools/kaldi/utils/int2sym.pl -f 2- /share/speech-to-phrase/train/en_US-rhasspy/graph/words.txt
DEBUG:speech_to_phrase.transcribe_kaldi:nbest: utt-1 office light to fifty 
utt-2 office light fifty 
utt-3 set office light to fifty 
DEBUG:speech_to_phrase.speech_tools:fstcompile | fstcompose - /share/speech-to-phrase/train/en_US-rhasspy/data/lang/G.fuzzy.fst | fstshortestpath | fstrmepsilon | fsttopsort | fstproject --project_type=output | fstprint --osymbols=/share/speech-to-phrase/train/en_US-rhasspy/data/lang/words.txt
DEBUG:root:Got transcription in 0.13302552001550794 second(s): 
DEBUG:speech_to_phrase.transcribe_kaldi:online2-cli-nnet3-decode-faster ['--config=/data/models/en_US-rhasspy/model/online/conf/online.conf', '--max-active=7000', '--lattice-beam=8.0', '--acoustic-scale=1.0', '--beam=24.0', '/data/models/en_US-rhasspy/model/model/final.mdl', '/share/speech-to-phrase/train/en_US-rhasspy/graph/HCLG.fst', '/share/speech-to-phrase/train/en_US-rhasspy/graph/words.txt', 'ark:/tmp/tmpcrbsj3le']
online2-cli-nnet3-decode-faster --config=/data/models/en_US-rhasspy/model/online/conf/online.conf --max-active=7000 --lattice-beam=8.0 --acoustic-scale=1.0 --beam=24.0 /data/models/en_US-rhasspy/model/model/final.mdl /share/speech-to-phrase/train/en_US-rhasspy/graph/HCLG.fst /share/speech-to-phrase/train/en_US-rhasspy/graph/words.txt ark:/tmp/tmpcrbsj3le 
LOG (online2-cli-nnet3-decode-faster[5.5]:ComputeDerivedVars():ivector/ivector-extractor.cc:183) Computing derived variables for iVector extractor
LOG (online2-cli-nnet3-decode-faster[5.5]:ComputeDerivedVars():ivector/ivector-extractor.cc:204) Done.
LOG (online2-cli-nnet3-decode-faster[5.5]:RemoveOrphanNodes():nnet3/nnet-nnet.cc:948) Removed 1 orphan nodes.
LOG (online2-cli-nnet3-decode-faster[5.5]:RemoveOrphanComponents():nnet3/nnet-nnet.cc:847) Removing 2 orphan components.
LOG (online2-cli-nnet3-decode-faster[5.5]:Collapse():nnet3/nnet-utils.cc:1488) Added 1 components, removed 2
LOG (online2-cli-nnet3-decode-faster[5.5]:CompileLooped():nnet3/nnet-compile-looped.cc:345) Spent 0.0130661 seconds in looped compilation.
2

I haven’t really played around with STP yet, but my understanding was that the matching needs to be pretty much exact. Your spoken phrase “set office light to fifty” doesn’t include the words “brightness” or “percent” as used in the built-in phrases.

3

When using STP but entering the command via text, it takes that happily. I’m not over-exposing entities either, so I can only conclude that it doesn’t need either brightness or percent to know what to do with the bulb. The bulb only has percent, it doesn’t have colour or any other attribute like that.

4

I’m not using STP either, but just going through a thought process … when entering text manually, it is probably bypassing STP and using the regular built-in intents, and looks like there is a built-in sentence for “set office light to fifty”. But as Didgeridrew mentioned above, STP is trying to use its own built-in phrases to match up with the voice/audio, and I too don’t see a STP built-in for “set office light to fifty”

5

Ah. I see. Hmmm… looks like I’m going to have to find more reading on this. It seems that the videos, etc. are a bit misleading. Making it look all nice and easy where there’s actually a lot more work to do.