Re: New music-based audio metric explained
Reply #2 – 2019-08-18 11:57:17
Such correlation can be discovered researching results of already finished listening tests. I made the preliminary/draft research based on results of HA listening tests - http://soundexpert.org/articles/-/blogs/objective-difference-measurements-to-predict-listening-test-results- Some conclusions: 1. The relation is pretty discernible. Better conclusions require examination of more strictly designed listening tests. And more of them. Such tests must be properly documented and the stimuli used have to be available. At the moment I have three more cases for research. If you know such well-documented and publicly available listening tests - please, share. 2. It turned out that this research requires a lot of time and hard work (approx. 6-12 months), and I'm afraid I can't complete it alone. I need help in the form of grant or something like this. If I perform this research with my usual schedule, it will take ages. 3. At this stage of the research the method of comparison Df levels with subjective scores is well-defined and first evidences of the correlation are revealed. Now the main task is to process as many listening tests' results as possible. Also some indirect evidences of the target correlation are in objective df-measurements of portable players and BT codecs - median Df values are consistent with intuitively graded perceived quality of the devices/codecs.