DeepSpeech MultiLanguage Server

Archives DeepSpeech
1

Good morning,

I have a basic question about using DeepSpeech.

I have the following project. I am setting up an Ubuntu 20.04 LTS server (headless) and want to install DeepSpeech there.
The task of the server should be to send audio files (vmtl ogg, opus) from a program via network to the server and let it process them. I do not know which language the audio files contain. As return I want to have the recognized text. I want to continue working with this in the program.

How can I implement this? The program I use to deliver the data has a Python interface.

Do I have to download different language models, if so where can I find them?

Greetings from Berlin
Bruno

2

DeepSpeech does not recognize the language (de/en/…) for you, maybe have several models transcribe your audio and see what gets good confidence values?

Please search before you ask, have a look at this discourse and the release and you’ll quickly find all available models.

DeepSpeech has Python bindings, so you can easily load several models. Look at the docs and the examples.

3

Hello @othiele !
Thanks for your input.

I will look @ the links.