Enhancement request: mixing with another repository and it's functionality

[mxav1111]

Hi there,
Thanks for the wonderful and clean front-end. Great work.
I did similar work in https://github.com/mxav1111/pdf_text_audio but through command line (parameterized through environment variables just as yours). While thinking about creating front-end , I stumbled upon this.
Front-end isn't really my forte. But if you are ok, we can mix match these both repository and provide all those options on front-end to generate audio (from scanned pdf), it would be great. I can talk to couple of other friends who are involved and we can put full working front-end with these all features (the ones you already have and the ones we will need to add front-end for options that are on command line) on NGO websites working extensively on literature and scriptures.
Microsoft based apis for pdf - text - mp3/wav, along with scripting for adding background music is also completed. Will upload shortly on pdf_text_audio repository. Everything in pdf_txt_audio is command line based and uses environment variables for everything.
PDF to TXT requires some involvement due to the fact that after text file creation, we need to perform manual correction including problems posed (especially by vision apis) .
Haven't spent much time on documentation but hopefully we can mix match this?

This seem to have great potential. In addition to options that are covered by environment variables, we also need to add few more such as emotion selection (if available for selected language), upload of pdf, download of text, upload of corrected text file, generating wav or mp3 , choice for quality , choice of background music , whether to split as well as merge of audio files and finally download of all this.
Website is just recently created with LOTs to do .. you can see some of the audio created from existing pdfs at https://awgpsac.org/gp

We use signal and if possible, we can regroup there... still need to upload updated scripts. Will do shortly.

What do you think? Thanks again for wonderful work.

[kneelesh48]

Hi,

I checked out your repo. The first thing I noticed is it's missing a requirements.txt file
I use environment variables but only for secrets like API keys. Using a .env file is convenient than typing all the environment variables in the terminal. Better than adding those commands to .bashrc because we don't want these environment variables outside the project.

You seem to have set the commands for Windows. But if we're going to publish it on some server or docker, it'll use linux. It's best to be OS agnostic as much as possible.
Add steps for initial setup and steps to run once initial setup is done in the README.
Making these changes will help me run this repo easily without digging into the code as I use a mac.

I'd like to keep this repo separate. It's a demo of all AI voices available on GCP and Azure.
I can help you create a frontend for this project and deploy on a public site. We can connect on Signal.

PS - I noticed you're using Tesseract OCR for OCR. I have a Tesseract OCR demo as well hosted here https://huggingface.co/spaces/kneelesh48/Tesseract-OCR

[mxav1111]

Thanks so much for offering to create front-end. That would be very very helpful.
Alright, so let me start with cleanup and setup the environment properly (and upload all the updated stuff etc..)
I was just lazy so set this all up on windows itself. I will now set this up on unix server and test with proper requirement.txt and shall update accordingly.
As for Tesseract, i was using it earlier but lately moved to google vision and MS document intelligence libs.
There's definitely quality difference especially when scan quality of pages (of very old books) are poor.
But we might need to jump back to it :-). Will check soon that too.

Here's signal group :- https://signal.group/#CjQKINo5oJkmktxaMqT3-TezPMJ0IF818vPK51nE51-sHuKNEhAlkxAwLXp8kdzlAJHG_7-Q

Thanks for your help.

[mxav1111]

ok. Successfully tested on Linux (debian). https://github.com/mxav1111/pdf_text_audio is updated.
using pip freeze (on windows) and used it on linux and after removal of some windows based libraries, requirements_linux.txt is created and tested.

https://awgpsac.org/gp/Bina_Ausadhiyo_Ke_Kayakalp.pdf was used and created https://awgpsac.org/gp/Bina_Ausadhiyo_Ke_Kayakalp.txt as well as https://awgpsac.org/gp/Bina_Ausadhiyo_Ke_Kayakalp_hi-IN-Neural2-A_0.85_m.mp3 on linux.

Thanks again for your interest. Hope that you will have bandwidth to be able to provide front-end.

[kneelesh48]

I've requested to join the Signal group. We'll have further discussion there.