Shop OBEX P1 Docs P2 Docs Learn Events
How to / Can the "mechanical / metal clacking " be filtered from this video's audio track? — Parallax Forums

How to / Can the "mechanical / metal clacking " be filtered from this video's audio track?

I'd like to remove as much as the "mechanical /metal clacking" as possible from
this video's audio track, leaving only the "morse tone", which is ~780hz



I have Audacity and Goldwave, any other program suggestions are welcome.

I tried messing with Audacity's "noise reduction" effect, and had some limited success,
but as an Audacity newbie, I'm thinking there might be better techniques ?

Comments

  • kwinnkwinn Posts: 8,697
    Try a notch filter at 700Hz with a Q of 20. After that you can copy and paste the track, invert the copy, subtract 714us (or as close as possible) from the front of the new track, then mix and render the two tracks to a new track. That new track should have a lot less noise.
  • Aren't the clicks the dits?
  • ercoerco Posts: 20,254
    xanadu wrote: »
    Aren't the clicks the dits?

    No, Alicia Silverstone is the ditz.


  • kwinnkwinn Posts: 8,697
    xanadu wrote: »
    Aren't the clicks the dits?

    No, that's mechanical noise from the keyer. the high pitched tones are the dits.
  • I guess I'm the ditz lol, all of the the tones sound the same length to me.
  • I'm pretty sure she sends dits with her with her thumb and dahs with her trigger finger.

    Check out the frequency counter at the top of the center rack.
  • Heater.Heater. Posts: 21,230
    Yeah, she has one of those trick Morse paddle things, hit it one way and it send out a string of "dits", hit it the other way and it sends out a string of "das". Fast.

    An old friend of my was really into Morse. One time his wife complained that he was tapping out Morse on the side of the bed whilst he was dreaming.

    A few years back I was dosing off in front of the TV only to be woken by the sound of ... --- ... coming from the TV, then a bunch of Morse code. Turned out this is part of some national emergency system in Finland. They were alerting us that a bear had walked out of the forest into a town some hundreds of kilometers away.

    A bit later I was woken up by more Morse from the TV. The all clear. The bear had been shot. Phew, now I can get some sleep. Poor bear.

  • Heater. wrote: »

    A few years back I was dosing off in front of the TV only to be woken by the sound of ... --- ... coming from the TV
    First thing we learned in Boy Scouts Morse Class, SOS.

  • Heater.Heater. Posts: 21,230
    edited 2016-05-02 17:25
    I'm never going to forget the old classics,
    -.-. --.-   -.-. --.-    -.-. --.-
    
    --... ...--    --- --
    

    Not that I was ever very good at Morse.

  • Heater. wrote: »
    I'm never going to forget the old classics,
    -.-. --.-   -.-. --.-    -.-. --.-
    

    Not that I was ever very good at Morse.

    Yup, second think we learned. :)
  • About half a century ago (1962/63), I was in a barracks with a squad of 058's (Manual Morse Intercept Operators). They used to go to the EM Club and drink lots of 3.2 beer. Then they would go to bed and carry on a conversation in Morse Code. Some would even vary the tone (as if the transmit frequency or BFO drifted). It was okay once I got used to it.
  • Try Reaper. It has a zillion plug-ins for audio processing.

    It was/is created by Justin Frankel, the creator of WinAmp who then sold it to AOL for like $millions. Free for 60 days and if you don't license it, it keeps working without restrictions anyway. At least this used to be true. I am a licensed user.
  • kwinn wrote: »
    Try a notch filter at 700Hz with a Q of 20. After that you can copy and paste the track, invert the copy, subtract 714us (or as close as possible) from the front of the new track, then mix and render the two tracks to a new track. That new track should have a lot less noise.

    For some reason the notch filter didn't work at all, but when I manually "drew" the envelope in the equalizer I got decent results.

    I also downloaded an "ultra narrow" filter from the Audacity forums which gave similar results to the equalizer

    @xanadu :
    @xanadu, I'm a beginner at morse, I've found a few good programs to learn it.

    1) http://morsecode.scphillips.com/translator.html
    Change the settings to 20 wpm 700 hz
    Type in cq cq cq de .. then I'm not sure .. sounds like TAPH/TST or TAPHTST

    Change the speed down to about 10 or 12, and space out the letters "C Q", you should start
    to hear the patterns.

    This is free neat program that lets you simualate *sending* morse using either your keyboard (down arrow)
    or mouse lmb/rmb (with the mouse can use iambic keying).
    http://www.mrx.com.au/d_cwcom.htm cwcom - can send with down arrow or mouse

    This is another good free program that will *read* incoming morse tones on your soundcard, as well
    as let you send ("text-to-morse typed in)

    http://www.qsl.net/hamscope/

    As for the message, it starts out with CQ CQ CQ DE then it sounds to me like either TAPHTST or TAPH/TST ?
  • ercoerco Posts: 20,254
    edited 2016-05-03 00:48
    Heater. wrote: »
    The bear had been shot. Phew, now I can get some sleep. Poor bear.

    This sounds like the punch line to my convoluted brain teaser in the "Speed of Light" thread.

    Yup, that wild woman is using a VibroPlex keyer, AKA 'bug'. I enjoyed my WB4VVF AccuKeyer in my formative years.

  • Yeah, she has one of those trick Morse paddle things, hit it one way and it send out a string of "dits", hit it the other way and it sends out a string of "das". Fast.

    They're called Iambic keyers. Squeeze one side you get dididididit, squeeze the other you get dahdahdahdah, squeeze both together and you get didahdidahdidah.
  • Actually Erco nailed it. That was a vibroplex bug. If you observe her index finger, it moves once for each dash, the dots are formed by a single thumb press. You varied the speed of the dots with a counter weight. Could not keep up with her, something about 60 years of no cw creating a lot of rust ....,.. As heater said --...,...--,---,--
    Jim
  • From a veteran cw'er on a ham forum I posted the same question, here is the morse message :

    cq cq cq de kph/kfs/ksm nw please join us for the traditional closing message bt

    dear goddess the members of the maritime radio historical society are your
    humble servants and we thank you for protecting us this past year as we
    continued our stewardship of the stations kph and ksm the music of morse has
    gladdened the hearts of many as we have crossed the barriers of time and space

    we ask your aid and guidance in our decisions and actions during the coming
    year that we may be worthy of the equipment and tradition that has been
    entrusted into our humble hands

    bless also the ears around the world that share
    the fruits of our labours zut 73/88 da de kph/kfs/ksm cl
  • Another thing you could try, if you just want the mores tones is to notch as suggested, then apply a limiter to fully suppress noise, leaving only the noisy Morse parts.

    Should you want a pure tone, normaize and compress what is left to get a brick wall.

    Then make a file full of the pure tone. Mix using the and function, and it will subtract, leaving you with a very pure tone with hints of the orginal signal.

    At that point, it's not original audio anymorr, but it will have the timings and hints of the original in it.
  • I got the impression that the OP just wanted the morse tone. Here is a little BS-2 program that will reproduce the tone, if not the operator's "fist". The audio comes out on I/O 1. You can adjust the speed with variable DotLeng and the tone with variable tone.

  • I just gave this a listen, and the morse tones aren't present on many "dits", due to the clack overwhelming the tone, and the tone itself having a bit of a ramp up to normal volume.

    That operator has a distinct timing too, and that's always of interest in most code discussions I've been a part of, as are the little pauses and variances in keying.

  • Heater.Heater. Posts: 21,230
    I thought the "mechanical clacking" and the twang of the springs in the VibroPlex keyer was the best part!

    The actual transmitted tones are something else.

    Denice Stoops is just so cool.
  • potatoheadpotatohead Posts: 10,254
    edited 2016-05-10 23:08
    Here's my effort, all original audio.

    https://www.dropbox.com/s/z0vqn7gkgyoqyds/ph-morse-tones.mp3?dl=0

    General approach:

    1. Normalize the audio

    2. High pass filter, then low pass filter, 48db aggressive. I used 730 and 830Hz as my starting boundaries.

    3. Apply notch filter, I used 30hz and the most aggressive settings.

    4. Use hard gate and Single Band EQ in order to pass maximized tones and drive the rest toward silent. (Gate limits by dB level)

    There is some ringing in it. Turns out I was wrong about the clacks. They do not completely overwhelm the morse tones. I probably could work with the gate more, filters less and get a bit better than what you hear on this one.

    Worth doing?






  • @Heater, this could be done again to isolate the clacking too. Then mix it back in at a reduced level.

  • Heater.Heater. Posts: 21,230
    She reminds me of a secretary that worked nearby my desk in 1990 something. She could type at a furious rate. The keyboard clacking made me crazy.

    Then they "upgraded" here computer from DOS and WordPerfect, or whatever, to Windows. She was immediately complaining like hell. The machine could not keep up with her typing speed.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2016-05-11 00:37
    Heater,

    A valiant effort. It's also an apt demonstration that frequency and time are conjugate variables. Narrowing the uncertainty in one (narrow-pass filter), increases the uncertainty in the other (ringing). There's no such thing as a free lunch in physics. It's the law! :)

    -Phil
  • Heater.Heater. Posts: 21,230
    Phil,
    A valiant effort. It's also an apt demonstration that frequency and time are conjugate variables. Narrowing the uncertainty in one (narrow-pass filter), increases the uncertainty in the other (ringing). There's no such thing as a free lunch in physics. It's the law!
    Do you mean Heisenberg's uncertainty principle?

    It is indeed the law.

  • Heater. wrote: »
    She reminds me of a secretary that worked nearby my desk in 1990 something. She could type at a furious rate. The keyboard clacking made me crazy.

    Then they "upgraded" here computer from DOS and WordPerfect, or whatever, to Windows. She was immediately complaining like hell. The machine could not keep up with her typing speed.

    My wife, back in the day, could jam up the ball on an IBM Selectric
    Jim
  • @Phil, that was me and a free 20 minutes, just wondering how good the stuff in Aucadity is. And it's OK. I've used much better and slower tools in the past. Still, if I were to give it another pass, a wider notch, one less iteration on that, and an aggressive gate would probably be clear enough for good code copy. And I do miss analog buttons, patches and dials... Could have managed the filter ringing much better by feel. Move the thingy, preview just sucks in comparison.

    IMHO, the best really would be be a mix down of the product of that second pass, and a file full of a pure tone. Brick wall compress the resulting file, then clip to make basically pure pulses out of it to preserve the timing, and the resulting mix would have the timing and subtle keying dynamics of the operator in it, done with an artificial tone. If one loads the optional Audacity plug-ins, a diode processor is there... Honestly, a lot of pretty good tools for freebies.

    I may do it, just for fun. The Aucadity tools are good enough.

Sign In or Register to comment.