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
7.36k stars 1.04k forks source link

Updating data to train an model #1523

Open gabrielroses15 opened 4 months ago

gabrielroses15 commented 4 months ago

Hey, what's up, bro? First, I would like to apologize about my English, I'm not an American, but I am trying my best. Well, my idea is taken a lot of audios in Brazilian Portuguese and transcript by myself with the best accuracy than I can, so after this, I want to put the audios and the transcriptions in a model to train it, exist any way as I put audios and they respective transcriptions to train a model? I'm asking this because The big model of FalaBrasil is good but not good enough, and I do not would like to start other model from 0 but, if I can, I prefer to train with my data, can I make this? If yes, how?

nshmyrev commented 4 months ago

Unfortunately you have to train the model from scratch if you need a better model. It is not very complicated.

We also have a telephony Portuguese model which is better than FalaBrasil, you can contact us with email and describe your project.