alphacep / vosk-api

Offline speech recognition API for Android, iOS, Raspberry Pi and servers with Python, Java, C# and Node
Apache License 2.0
7.71k stars 1.08k forks source link

Swedish model #686

Open Falsen opened 3 years ago

Falsen commented 3 years ago

This project really looks amazing! Are there any plans to create a swedish language model?

nshmyrev commented 2 years ago

Thank you. I believe you can use this model if you arrange the files properly:

https://github.com/rhasspy/sv_kaldi-rhasspy

there are big models too trained on Spraakbanken.

Do you need a big or a small model for mobile?

nshmyrev commented 2 years ago

Try https://alphacephei.com/vosk/models/vosk-model-small-sv-rhasspy-0.15.zip

Falsen commented 2 years ago

@nshmyrev Cool! I didn't find that one for some reason, will try it out! I need the best one I can get, and judging by the size it is quite large? Thank you!

nshmyrev commented 2 years ago

I need the best one I can get, and judging by the size it is quite large?

It is pretty small. We could have bigger one eventually.

Falsen commented 2 years ago

Okey, I tried it. Sadly it wasn't very accurate when testing. Are there parameters that can be tuned in any way to achieve a better result, or is the model size the limiting factor?

nshmyrev commented 2 years ago

Are there parameters that can be tuned in any way to achieve a better result

It depends on the application you want to build