search

stefanpantic / asr

Automatic speech recognition using neural networks
GNU General Public License v3.0
19 stars 1 forks source link

Is this a mistake, about QuartzNet model layer configuration #33

Closed hellomotor closed 3 years ago

hellomotor commented 3 years ago

hi, i think there's a mistake here, https://github.com/stefanpantic/asr/blob/86e13ee4d76779318caf66a81d6342c096e5fb35/models/quartznet.py#L30, should be: filters: [512, 1024], kernel_size: [87, 1], can you confirm it?

stefanpantic commented 3 years ago

Yes, they should be swapped, good catch!