khalian / modulo7 Goto Github PK

Once in production, music stat models will not be able to stay in memory as the number of songs processed by the crawling engine will go over the in memory capacity.

Need to work on good serialization strategies. Investigate hibernate and avro for these purposes (hibernate for object to relational mapping and avro for low latency seraialization)

Investigate apache J clouds as a client lib and openstack as a server env.

Complete the docker support for modulo7

Currently modulo7 dependency installation is complete on top of the docker image construction. Need to build docker image with modulo7 installed on it and scripts to exec modulo7 components from docker container.

Aakash Can you have a look at this?

As described, investigate Hbase, spark etc to figure out how to go about parallelized framework.

Investigate Key Estimation Algorithms

Need to remove dependency on Echo Nest for MP3 Analysis.

ECHO Nest applies a API Key and a rate limit. Cannot scale in full production and will be fully dependent on the company.

Acoustics Engine must support in house programmatic solution to MP3 analysis.

I noticed that musicXML format also includes lyrics. Good chance to extract lyrics i guess?

Check if we need to support humdrum or not. It would depend on the ubiquity of the format.

http://www.musiccog.ohio-state.edu/Humdrum/

Queries return a set of songs that satisfy a criterion or statistic to some value

To add support for subquery so that support can extend to multiple nesting of queries and boolean logic to add them together.

I noticed that musicXML format also includes lyrics. Good chance to extract lyrics i guess?

Add support for music xml.