Any Good Architecture for continous Inference

hnipun · February 12, 2018, 9:24am

Hi,

Is there any good resources to learn to implement a continuous speech to text system using. From microphone sound capturing, segmenting the continuous sound signal and any preprocessing methods till feed the DeepSpeech model.

Thanks in advance.

lissyx · February 12, 2018, 9:44am

You can probably have a look at the Voice Fill WebExtension, it’s not relying on DeepSpeech, but some people have provided DeepSpeech “servers” implementation on Python and NodeJS.

Elleo · February 13, 2018, 1:20pm

You might like to take a look at the GStreamer DeepSpeech plugin I wrote, which does some basic segmentation (just based on silence thresholds):

http://blog.mikeasoft.com/2017/12/30/speech-recognition-mozillas-deepspeech-gstreamer-and-ibus/

hnipun · February 14, 2018, 4:18am

Thanks a lot. Its a really nice idea. Do you have any documentation so that I can learn about your architecture.