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
7.35k stars 1.04k forks source link

What does the new postprocessors do? #1562

Closed msqr1 closed 2 months ago

nshmyrev commented 2 months ago

They can convert numbers to digits like "five thousand" -> 5000. You can get corresponding models here:

https://github.com/NVIDIA/NeMo-text-processing

msqr1 commented 2 months ago

Is this integrates into Vosk so I have to compile them together, or this is a separate component?

nshmyrev commented 2 months ago

Its part of the vosk, no need for anything extra

msqr1 commented 2 months ago

So the asr is gonna output 5000, and this is going to convert it to five thousands? I can compile and run just fine without it, i dont see why I need it to postprocess the result?

nshmyrev commented 2 months ago

So the asr is gonna output 5000, and this is going to convert it to five thousands?

opposite way

I can compile and run just fine without it, i dont see why I need it to postprocess the result?

If you want to see 5000

msqr1 commented 2 months ago

Oh ok, thanks so much!