search

abhiandthetruth / perceval-voice

Enabling Voice Queries on Grimoire Labs Perceval
1 stars 0 forks source link

Basic Structure of the module #1

Open abhiandthetruth opened 4 years ago

abhiandthetruth commented 4 years ago

Commands can be like "Sir Perceval, please fetch me the information from the repository grimoirelab of organisation chaoss from github" or something like this. We may remove the "Please" though ;). This would be an attractive feature for Grimoire Lab. I know that there are several challenges like 1) Nlp requirement if we allow voice commands directly from natural languages (eg. English) 2) Development or adoption of an unambiguous query language if we do not use a natural language and instead opt for a syntactic form of English with predefined functions and query params. 3) The names of repositories and usernames are difficult to convert from speech to text with good accuracy. This may itself require search query for nearest matches.

There are more of them which would be exposed during the phase of development.

So the basic questions are: 1) Are we using natural language or syntactic one as input? We can also move up like first implementing it with a syntactic language and then implementing nlp? 2) What are the libraries or apis we will be using? You have better knowledge than me on these issue @valeriocos. What are your decisions on the above questions?

abhiandthetruth commented 4 years ago

We can have a backend in flask and a minimal frontend in html/js to take the voice input and display the results processed by the back end running perceval.

valeriocos commented 4 years ago

Sorry for the late reply @abhiandthetruth and thank you to kickstart the discussion!

As a general comment, I would go for the easiest solution to get a very simple PoC and start diving into the technical difficulties later. Thus, a minimal frontend is perfect at this stage. For point 1. I guess a syntactic language may be better (please share any useful reference). WRT point 2. I need more info about the libraries or APIs you refer to. Can you elaborate more on this?

abhiandthetruth commented 4 years ago

@valeriocos In the second point I was talking about using libraries for speech to text conversion. I wanted to know if you knew about some good open source one which we can use. Using Google's API would not be good for an open source project I guess. I will be researching on the syntactic language we could use for perceval keeping in mind perceval's api. Meanwhile can you please look into the speech to text issue.

abhiandthetruth commented 4 years ago

Also regarding perceval repo, while doing the micro-tasks I noticed that there were no example codes highlighting perceval's api. Shall I work on some descriptive sample codes covering the API methods and classes and make a pr. It would be helpful and will attract new developers to try perceval :). WDYT @valeriocos ?

valeriocos commented 4 years ago

https://github.com/abhiandthetruth/perceval-voice/issues/1#issuecomment-592945326

ok! I won't be able to do it until Tuesday, since I don't have my laptop with me. Sorry for the inconvenience.

valeriocos commented 4 years ago

https://github.com/abhiandthetruth/perceval-voice/issues/1#issuecomment-592945703

good idea! there is already some doc at https://chaoss.github.io/grimoirelab-tutorial/contributing.html#. Feel free to expand/modify it, thanks :)