search

alphacep / vosk-api

Offline speech recognition API for Android, iOS, Raspberry Pi and servers with Python, Java, C# and Node
Apache License 2.0
8.12k stars 1.12k forks source link

Updating the model, there are no output symbols #1342

Open Kalla92 opened 1 year ago

Kalla92 commented 1 year ago

Hi,

in my everyday work-routine I have some repetitive tasks, which I need go through many times. So I would like to automate it with some speech commands.

I found this model (models/vosk-model-small-en-us-zamia-0.5(49M)) suitable for my purpose, but I wanted to improve the recognition of my specific commands(words). So I followed this steps: https://alphacephei.com/vosk/adaptation , the "Updating the language model" section. But while building the grammar I ended up with this error: "... fstsymbols: Saving osymbols but there are no output symbols". My text.txt contains only these 3 lines for example(these aren't new words for this model): resistor mega kilo

Is it possible to make modification on this model like this? (I was able to build the grammar on another model: vosk-model-en-us-0.22-lgraph(128M), but this model doesn't works well for me). Thank you for any advice.

nshmyrev commented 1 year ago

You can use words.txt from the model as a symbol table.

amina1403 commented 1 year ago

You can use words.txt from the model as a symbol table.

Is there a way to increase the accuracy?