3 ms·
Probably positively. I brought up DeepSpeech in a docker container the other day, it couldn't understand me at all, using Mozilla's pre-trained model. _Some_ wo
by m0zg 8y ago
Probably positively. I brought up DeepSpeech in a docker container the other day, it couldn't understand me at all, using Mozilla's pre-trained model. _Some_ words were right, but the output did not make sense.
- StudentStuff 8y agoDid you use one of the examples (eg the Python GUI or CLI client)? WebRTCVad (which is used in most of those) made mincemeat of my audio files, versus feeding them into DeepSpeech got me usable results. Just need VAD for files nearing or over 1 minute in length (unless you have 4+GB of RAM to spare).