MythMusic: Improve Spectrum and Squares

[twitham1]

This applies the lessons learned in Spectrogram to the original Spectrum and Squares, greatly improving them:

• FFT of complete signal
• same FFT: 16384 bins, decaying window, peak of bins
• same Mel Scale replacing log scale (much improved resolution in low frequencies, distributes the sound better across the whole screen)
• upgrade to 128 bars for better resolution

[Jpilk]

I have master 0279fd2 running now, with MCW1.10 locally patched to display recordedid as in issue #464.

The only difficulty seen so far is inability to quit full-screen visualiser when the ident popup is continually refreshed, as with ClassicFM streaming. That needs ctrl/c to quit mythfrontend.

http://ice-sov.musicradio.com:80/ClassicFMMP3

[linuxdude42]

A couple of minor comments.

[linuxdude42]

Possibly replace this with something like:

m_analyzerBarWidth =
    std::max(m_size.width() / kNumBarsWanted, kMinBarSize);

and define the two constants in the Spectrum class with:

static constexpr int kNumBarsWanted {128};
static constexpr int kMinBarSize {6};

I know there are close to 15,000 other magic numbers in the code, so I'll leave it to you whether to change this. I do think this type of change makes the code more readable.

[twitham1]

Personally, for a number used in 1 place, the number is easier to read than words. Because if it is words I have to go lookup where the words resolve to a number. So reading 128 and 6 right here is easier for me.

But certainly numbers used in more than 1 place should be words instead.

[linuxdude42]

You understand the code, so its not surprising that you find the numbers easier. For someone reading the code for the first time, the descriptive constant names would be easier.

[twitham1]

The only difficulty seen so far is inability to quit full-screen visualiser when the ident popup is continually refreshed, as with ClassicFM streaming. That needs ctrl/c to quit mythfrontend.

http://ice-sov.musicradio.com:80/ClassicFMMP3

Wow, that's bad! But totally unrelated to this PR. I think I can fix it but let's work on that over on #790.