Results 1 to 3 of 3

Thread: Music matching of older "standard" songs just by different instrument played melody?

  1. Top | #1
    Contributor repoman's Avatar
    Join Date
    Aug 2001
    Location
    Seattle, WA
    Posts
    6,411
    Archived
    2,280
    Total Posts
    8,691
    Rep Power
    73

    Music matching of older "standard" songs just by different instrument played melody?

    I was watching some old episodes of Name That Tune, which I was a smidge too young to have watched when new, and I am impressed with the ability of the best contestants.

    It is not played on the same instrument normally, granted these standard songs were highly melody driven and hummable. In addition there were a some title clues.

    Let's say that you had a database of only hummable standards (this is where the rub lies) from about 1930-1980 played on piano or small band. Also, would it only be based on notes from the beginning of the song?

    What about slightly divergent covers of these songs?

    Would finding out how humans tell these songs apart make search algorithms faster?

  2. Top | #2
    Veteran Member
    Join Date
    Oct 2018
    Location
    Colorado
    Posts
    1,565
    Rep Power
    5
    We are about the same age, it seems. I had an electronic "Name that Tune" game. It had a very limited database of songs and it worked just like in the game show. MIDI versions of the melodys.
    OMG, I found it...

    Just noticed it says you can make your own songs on the box.. I don't remember that at all. Maybe I had an older version...


    I think you will find this TED talk very interesting... This guy did a novel analysis of pop music that I bet you will really like. An interesting fellow, he is. Not what you are going to initially expect. It is only somewhat on topic here, though...


    (View video on YouTube)

  3. Top | #3
    Veteran Member
    Join Date
    Nov 2017
    Location
    seattle
    Posts
    4,924
    Rep Power
    12
    Early jazz was about taking a standard and improving variations.

    I tend to recognize people by voice before face. The primary process appears to be frequency decomposition. Or Fourier Transforms. We compare sounds without having to think about it.

    Analyzing a song for a beat and melody doesn't sound that difficult. Once the analysis parameters of the analysis is set then it is pattern recognition.

    It sounds like something suited for neural nets, pattern recognition.

    One of my favorites is Ramsey Lewis's piano jazz version of the pop song The In Crowd.

    https://www.bing.com/videos/search?q...BF0F&FORM=VIRE

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •