Prerequisite Dependencies

1. node-tesseract-ocr

• Tesseract

	aptitude install tesseract-ocr

2. 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

3. @samuelcalegari/ds_ffmpeg, ffmpeg-extract-audio, ffmpeg-extract-frames-quality

• ffmpeg

	aptitude install ffmpeg

---

Heroku Instructions

1. Create heroku application

heroku create <name>

2. 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

3. 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.

4. 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).