Will the results be different when use the trained model under "deepspeech" and "deepspeech-gpu"?


(jackhuang) #1

I can’t use the “deepspeech-gpu”, so I use the “deepspeech” to get the result. Will the results be different when use the trained model under “deepspeech” and “deepspeech-gpu”?


(Lissyx) #2

There should be no difference at all, except of course the inference time :slight_smile:


(Megha ) #3

How did you write a command to use deepspeech-gpu? I tried just deepspeech to get the result. it works, but I want to try with deepspeech-gpu. can you share your gpu based commands to get the output?

Thanks :slight_smile:


(Lissyx) #4

There is no difference. Just make sure you uninstall deepspeech python wheel and install deepspeech-gpu instead.


(Megha ) #5

Hi @lissyx,

Do you mean like below?

pip uninstall deepspeech
pip install deepspeech-gpu

if so, When I do

deepspeech-gpu models/output_graph.pb my_audio_file.wav models/alphabet.txt

I am getting

deepspeech-gpu: command not found

But with pip list I am able to see deepspeech-gpu in my system.

any suggestions?

Thanks :slight_smile:


(Lissyx) #6

This is all documented in the README.