mikrosimage / loudness_validator Goto Github PK

View Code? Open in Web Editor NEW

13.0 13.0 8.0 754 KB

A set of applications to analyse, visualize and correct the loudness.

License: Other

Python 9.28% C++ 89.76% QMake 0.04% CSS 0.03% Shell 0.89%

loudness_validator's People

Contributors

Stargazers

Watchers

Forkers

cchampet fingul marcantoine-arnaud adoussot 4lm media-io jon68z inoculate23

loudness_validator's Issues

Use SSE2 instructions to optimize filter process

https://github.com/mikrosimage/loudness_validator/blob/develop/src/loudnessAnalyser/Filter.cpp#L81

Add unit tests for process

filters
with and without SSE2

Cannot find qt4 when try to build loudness-validator on MacOS X

But the link works when I add 'not' to the following 'if':
https://github.com/mikrosimage/loudness_validator/blob/develop/app/SConscript#L54

Deploy new linux build to Github release

For each new version, deploy the build to the Github tag.

Update LoudnessValidator app: switch to a web app

No more Qt.

Test: fix check of the momentary loudness

https://github.com/mikrosimage/loudness_validator/blob/develop/test/loudness-analyser.cpp#L208
https://github.com/mikrosimage/loudness_validator/blob/develop/test/loudness-analyser.cpp#L214

use SCons Options (to display on help)

refers to 12.1.5 http://www.scons.org/doc/1.0.0/HTML/scons-user/c1965.html
Change our ARGUMENTS, gets in SConstruct (https://github.com/mikrosimage/loudness_validator/blob/develop/SConstruct#L18)

adding aiff export

like the Wav format, the application need to correct also the AIFF format.

Can analyse loudness between given timecodes

madia-loudness-analyser: possibility to analyse one media essence, with multi programs (Stereo, Multichannel, VO/VF/AD/etc.) using only one process.

Short Term Limit for short program

According the current specification, R-128 expects a short term limit of -20 LUFS without tolerence (for short programs).
https://github.com/mikrosimage/loudness_validator/blob/develop/src/loudnessAnalyser/LoudnessAnalyser.hpp#L140

Split project in 3 loudness libs: analyser / corrector / Tools(io)

No need to compile loudnessTools library if we can access some audio buffer from elsewhere.

Add unit tests (C++) of loudness analysis

With new zipfile (v4.0 of the test set) from EBU: https://tech.ebu.ch/publications/ebu_loudness_test_set

Fix analyse of loudness with standard ATSC_A85

Add SSE2 flags (automatic detection or force)

Could increase perfs x8
http://softpixel.com/~cwright/programming/simd/cpuid.php
http://stackoverflow.com/questions/13757742/cpuid-instruction-on-i5-2500k-mmx-sse-sse2-bits-arenot-set
https://software.intel.com/sites/default/files/c8/ab/17971intel_20sse4_20programming_20reference.pdf
https://msdn.microsoft.com/en-us/library/hskdteyh.aspx

Generate java binding

Use SWIG library.

Optimize the analyse process

These following points can be a started point:

Why did it divide by 5? Could the buffer be larger than that?
https://github.com/mikrosimage/loudness_validator/blob/develop/src/tool/common/ProcessFile.hpp#L12

Could this loop of affectations be avoided?
https://github.com/mikrosimage/loudness_validator/blob/develop/src/tool/common/ProcessFile.hpp#L31

Could this call be multi threaded?
https://github.com/mikrosimage/loudness_validator/blob/develop/src/loudnessAnalyser/LoudnessAnalyser.cpp#L35

@ChristopheLorenz
@MarcAntoine-Arnaud
Do you have any idea of how to improve performances of analyse process?

Add test coverage report

LoudnessAnalyser::processSamples: why the audio buffer contains float

loudness_validator/src/loudnessAnalyser/LoudnessAnalyser.hpp

Line 218 in 30829c8

void processSamples( float** samplesData, const size_t nbSamples );

Is it because libsndfile returns an array of float, or because of the loudness algorithm?

Optimization: skip copy of audio buffer before each process

https://github.com/mikrosimage/loudness_validator/blob/develop/src/tool/common/ProcessFile.hpp#L54
This involves copy of data, and caching problems.

Can analyse audio from a given buffer

Instead of reading data from a file: https://github.com/mikrosimage/loudness_validator/blob/develop/src/tool/common/ProcessFile.hpp#L97

Setup CI (use Travis)

Create & push homebrew formula to homebrew

create rule for homebrew
create debian packaging