I have a collection of live recordings of very varying sound quality.

I have graded them in a first pass (along with tagging them) but I'm afraid that evaluation was rather inconsistent.

The grades went from A++ to C but I felt during the process that I should find a more objective and more detailed evaluation system.

So I have started writing a second tool to do just that. It shall play randon portions from each show and I would fill in/check a number of parameters.

Here is a screenshot of what it looks like at the moment:

enter image description here

The buttons are actually radiobuttons which will display a color corresponding to the weight (+3 to -3) of the parameter. (The last row buttons are checkboxes, though, to allow checking more than one type of audience 'noise'.) I then will store both the details and a summary grade.

My question is:

  • What suggestions come to your mind?
  • What is missing?
  • What better terms should I use?
  • Is there something like this out there already?
  • Can I please ask, what is the purpose of the grading exercise? What you have described is a solution to a problem, but I am not sure what the problem is yet. – Mark Feb 27 at 14:10
  • The purpose is nothing unusual I'd say: I want to know which of the downloads should go into the inner circle of music to keep online in my home collection. I download what seems to be interesting but many recordings are not good enough to actually listen to. Also a somewhat reliable grade is helpful when setting up trades. The full body of shows is around 2-3k, so some system is called for.. – user26499 Feb 27 at 14:24
  • 1
    Ok understood. As per answer below, I would probably prioritize technical quality over everything else. No point in having a great mix that you just can't listen to because the recording sucks. – Mark Feb 27 at 14:27
  • 2
    I don't feel that this is anywhere near useful to anyone but you. There are endless records and mixes that would score oddly in the chart you've produced that are godly, (e.g. Rhythm and Sound records would be nothing if the hiss, buzz, detune, bad shaped delay repro/record heads and on). So in any way this is not an objective representation of good or bad sound/mix. I'd stick to the first evaluation. Otherwise just run some pink noise and score it. That said, that's my personal opinion on the subject, if it helps you, it's cool! I just find it too technical VS the psychoacoustic music effect – frcake Mar 1 at 14:59
  • This reminds me of a project I did in University. But it was to do with percieved quality of file compression algorithms for different genres. There was only one parameter though. "Sound Quality" which was intentionally subjective. Great fun. – Marc W Apr 4 at 15:24

I would approach this looking at objective criteria such as:

  1. Mix Balance - are the elements of the performance correctly balanced? For instance do the vocals sit in the right place in the mix along with the rest of the instrumentation?
  2. Spectral Balance - is the mix a full-range mix or are there elements of the spectrum that are attenuated?
  3. Technical Quality
    • does the recording suffer from artefacts such as encoding noise, preamp noise, disc/stylus malformation etc?, print-through (tape), etc.
    • is the phase relationship between the channels correct?
    • what is the stereo spread like? Too wide, ok, or mono?
  • Yes, 'objectvie' data are also quite welcome, although I'm not sure how to a) get them or b) how to interpret what I get. I'm looking into software for spectrum analysis. Some efficiency is also called for as I have a few thousand live recodings to go through, so the checklist approach seems promising; the question was how to improve the checklist with more or better criteria.. – user26499 Feb 27 at 14:27
  • 1
    I honestly think that the only solution is ears. It will take a while, but I think it's the right way to go. There's no technical or analysis solution that can rate a recording in this way. – Mark Feb 27 at 14:29
  • Agreed. But ears alone tend to get tired, so breaking it down to simple questions could help.. – user26499 Feb 27 at 14:30
  • @TaW oh definitely, breaking it down into simple questions is a good idea for sure, but I would approach it along the lines of the answer above. Maybe a 1-5 rating for each would be enough. – Mark Feb 27 at 21:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy