The persephone ASR project is an open source tool which allows you automate speech-to-phoneme transcriptions. It is built with a machine learning engine that can be trained on your own labelled audio test data and has been successfully used with many spoken languages worldwide. Due the the ability to get results from small sets of labeled training data it has found use in the study of low resource languages. (See this paper for more info) It can run on premise on your own machines and is capable of running on both CPU and GPU hardware.
PEP 479 changed how Python handled generators in a backwards incompatible manner, here's how to get compatible code working from Python 3.5+Published on July 3rd, 2019 by Janis Lesinskis.
I was bumping some versions for the Persephone project today to try remove some security issues from the project dependencies. Package…Published on June 28th, 2019 by Janis Lesinskis.