bnriiitb / chitralekha-backend

Transcribe your videos and translate it into Indic languages.

Chitralekha Backend

Transcribe your videos and translate it into Indic languages.

You can watch a demo of our tool - Video

Setting up the Speech server locally

1. Clone the repository - git clone --recursive https://github.com/AI4Bharat/chitralekha-backend.git
2. Change directory - cd pratilipi/speech/api
3. Make sure you have torch > 1.7 installed
4. Install fairseq,kenlm,flashlight bash setup.sh
5. Install dependencies by running pip install -r requirements.txt
6. Download acoustic and language models -

wget https://storage.googleapis.com/indicwav2vec-public/shared_to_vivek/evW2V_12K_64.pt
wget https://storage.googleapis.com/indicwav2vec-public/shared_to_vivek/lexicon.lst
wget https://storage.googleapis.com/indicwav2vec-public/shared_to_vivek/lm.binary

8. Change model path in test.json
9. Run the server using python flask_api.py <cpu/cuda:0> uvicorn main:app --host 0.0.0.0 --port 5000

Setting up the Translation server locally

1. Clone the repository - git clone --recursive https://github.com/AI4Bharat/pratilipi.git
2. Install dependencies - bash download_and_setup_translation_models.sh
3. Fire up the flask server locally - FLASK_APP=api flask run -p 5050 -h 0.0.0.0 uvicorn main:app --host 0.0.0.0 --port 5050

Setting up the Feedback API server locally

1. Change to the feedback api directory - cd feedback-api
2. Install dependencies - pip install -r requirements.txt
3. Fire up the flask server locally - FLASK_APP=api flask run -p 6070 -h 0.0.0.0

Setting up the Frontend server locally

1. Clone the repository - git clone --recursive https://github.com/AI4Bharat/pratilipi.git
2. Make sure you are using a python version >= 3.5
3. Change the api http paths in the following lines in index.html. This step should be done once all the api servers are up and running.

Line numbers - 408, 409, 410

4. Fire up the HTTP server locally - python -m http.server 8090 -b 0.0.0.0