Transcribe your videos and translate it into Indic languages.
You can watch a demo of our tool - Video
- Clone the repository -
git clone --recursive https://github.com/AI4Bharat/chitralekha-backend.git
- Change directory -
cd pratilipi/speech/api
- Make sure you have torch > 1.7 installed
- Install fairseq,kenlm,flashlight
bash setup.sh
- Install dependencies by running
pip install -r requirements.txt
- 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
- Change model path in test.json
- Run the server using
python flask_api.py <cpu/cuda:0>
uvicorn main:app --host 0.0.0.0 --port 5000
- Clone the repository -
git clone --recursive https://github.com/AI4Bharat/pratilipi.git
- Install dependencies -
bash download_and_setup_translation_models.sh
- 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
- Change to the feedback api directory -
cd feedback-api
- Install dependencies -
pip install -r requirements.txt
- Fire up the flask server locally -
FLASK_APP=api flask run -p 6070 -h 0.0.0.0
- Clone the repository -
git clone --recursive https://github.com/AI4Bharat/pratilipi.git
- Make sure you are using a python version >= 3.5
- 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
- Fire up the HTTP server locally -
python -m http.server 8090 -b 0.0.0.0