added the ability to use more than 3 files with backwards compatibility.

[hanzala123]

No description provided.

[NXture]

hey, nice work @hanzala123 , may i know the docker command which we can use over the default one?
python generate_pmdl.py -r1=record1.wav -r2=record2.wav -r3=record3.wav -lang=en -n=hotword.pmdl -> this is the current one, which takes only 3 set of wav files.

[hanzala123]

This is the command
python generate_pmdl.py -rD=DIR_CONTAINING_WAVS -lang=en -n=hotword.pmdl

[NXture]

Thanks for that @hanzala123
I was wondering that can we increase the hotword detection efficiency by using more than 3 record samples? Because Earlier i used 3 record samples and generated a hotword.pmdl file which was 8 KB, But now i used 25 samples, and the model size was 9 KB. So i just wanted to make sure that using multiple samples makes any difference?

[hanzala123]

@NXture Well i only used it to train one model which gave me better results than what i was having with three wav files.
Though this might not be the case for everytime. It needs more testing before i can confirm anything.