audio2text

docker-based pocketsphinx that takes A/V input file to create text from the audio and extract keywords and entities from the results.

Allows one to, once SETUP is done, run locally on a Mac, network disabled / no cloud, and run speech to text and text analysis on an A/V file.

PREREQUISTES

Docker (eg: https://docs.docker.com/docker-for-mac/install/ )

git (brew or XCode setups have you covered ;-)

SETUP

git clone https://github.com/traceypooh/audio2text.git
cd audio2text
docker build -t audio2text .

RUN

( docker run --rm -i audio2text |tar xf - ) < test.mp3

will make (click each to see the results):

• out.json - detailed word/phrase with timings
• out.txt - transcript of entire audio/video file
• out.srt - timed transcript of audio/video file
• out.key - keywords extracted from .txt (above)
• out.plo - Persons, Locations, Organizations (and more) extracted from .txt (above)