what Software can I use to convert an audio file with spoken german to a text? It should run under Linux, ideally FOSS and offline.
It should come pre-trained, I do not want to train it.
2 Answers
You can use Vosk, it supports German, runs offline and can transcribe speech with Python. You can also use the speech recognition server from docker.
Latest big German model is here (1 Gb):
https://alphacephei.com/vosk/models/vosk-model-de-tuda-0.4.zip
There is also a small one for lightweight applications.
I work for Keen Research, https://keenresearch.com, which develops and licenses SDKs for on-device speech recognition on iOS, Android (we support Linux also).
-
1Thanks for this answer and for disclosing your affiliation. Is there any webpage online about this product? I remove the email address per our policy.– Nicolas Raoul ♦Commented Jan 20, 2022 at 9:31
-
@NicolasRaoul Yes, here is the link to the trial SDK keenresearch.com/keenasr-docs/try-keenasr-sdk.html Commented Jan 21, 2022 at 2:46
-
From keenresearch.com/keenasr-docs/index.html : Speech recognition is performed locally on the device; no internet connectivity or cloud support is required– Nicolas Raoul ♦Commented Jan 22, 2022 at 8:32