Skip to content

Latest commit

 

History

History
64 lines (56 loc) · 1.51 KB

README.md

File metadata and controls

64 lines (56 loc) · 1.51 KB
Raw

Prerequisite Dependencies

  1. node-tesseract-ocr
  • Tesseract
	aptitude install tesseract-ocr
  1. deepspeech
  • Model and Scorer
	# Download pre-trained English model files
	curl -LO https://github.com/mozilla/DeepSpeech/releases/download/v0.8.1/deepspeech-0.8.1-models.pbmm
	curl -LO https://github.com/mozilla/DeepSpeech/releases/download/v0.8.1/deepspeech-0.8.1-models.scorer
  • Python3
	aptitude install python3
  • TensorFlow
	pip3 install tensorflow
  • Sox
	npm i sox
  1. @samuelcalegari/ds_ffmpeg, ffmpeg-extract-audio, ffmpeg-extract-frames-quality
  • ffmpeg
	aptitude install ffmpeg

Heroku Instructions

  1. Create heroku application
heroku create <name>
  1. Add buildpacks This is lot of buildpacks. I've tried at most 3. Be careful.
1. heroku/python
2. https://github.com/AirspaceTechnologies/tensorflow-buildpack.git
3. heroku-community/apt
4. https://github.com/jonathanong/heroku-buildpack-ffmpeg-latest.git
5. heroku/nodejs
  1. Deploy and test
git add .
git commit --allow-empty -m "Initial deploy"
git push heroku master

At this point a very long build process should run and download everything required.

  1. Note the size of the compressed slug (it'll come up at the end of the build process). If this size is less than or around, say, 250 MB, try enabling scorer and check if it fits. To enable scorer, go to .heroku/run.sh and uncomment the last (or second-last) line that correspons to scorer. Then, re-deploy (step-3).