build big US English model with dynamic graph

Hello, Nickolay V. Shmyrev

To generate a dynamic grammar based on the VOSK framework, I referred to [https://github.com/alphacep/vosk-api/blob/master/python/example/colab/vosk-adaptation.ipynb](https://github.com/alphacep/vosk-api/blob/master/python/example/colab/vosk-adaptation.ipynb) and added sentences containing proper nouns to db/extra.txt and the corresponding pronunciation dictionary to db/extra.dic. Then, within compile-graph.sh,
I added
`utils/mkgraph_lookahead.sh \
--self-loop-scale 1.0 data/lang \
exp/tdnn data/en-us-mix.lm.gz exp/tdnn/lgraph
utils/mkgraph_lookahead.sh \
--self-loop-scale 1.0 data/lang \
exp/tdnn data/en-us-mix.lm.gz exp/tdnn/lgraph`

and run ./compile-graph.sh

From the generated results, I copied the lgraph directory
into the graph within VOSK and tested sentences containing the newly added words using the dynamic graph, and the process proceeded without any issues.

I copied the `final.md` and `tree` files from the new model `vosk-model-en-us-0.22-lgraph` to `exp/final.md` and `tree` for the `compile-graph` build, and then ran `compile-graph.sh` as described above. Although no errors occurred, no recognition results were produced.
I would appreciate any guidance on how to proceed.

Thank you.