[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference napalm::commusic_v1

Title:* * Computer Music, MIDI, and Related Topics * *
Notice:Conference has been write-locked. Use new version.
Moderator:DYPSS1::SCHAFER
Created:Thu Feb 20 1986
Last Modified:Mon Aug 29 1994
Last Successful Update:Fri Jun 06 1997
Number of topics:2852
Total number of notes:33157

2700.0. "Slowing down speech and music " by 3D::ASFOUR () Wed Aug 21 1991 16:55

	I've posted this note in RUSURE::MATH.
	Any help and/or pointers would be a great help.

    I'm not sure which notesfile to post this in, so I figured
    I'd start here.
    
    I've scanned a couple of papers describing several methods to
    change the speed of speech without changing its spectral
    characteristics. 
    
    e.g. play back speech slower without making it sound lower in pitch
    or distorting it.
    
    Does anyone know anything about any such techniques? Does anyone
    know if these techniques (or their equivalents) would work equally 
    well with other forms of sound such as CD quality music? 
    
    I'm working on a multi-media A/D project and we would like to be able 
    to synchronize several sources of audio and video together. Sometimes,
    the audio could take a longer or shorter time than the video,and we
    would like to be able to "shrink" or "expand" it so that they would 
    fit.
    
    Any pointers to "cookbook" methods to implement these (as opposed
    to translating the math in these papers to programs) would be 
    a great help. 
    
    		thanks.
    			Yousif.
    
    P.s. The papers I'm refering to are found in ___Speech Enhancement__
    edited by Jae S. Lim, Prentice-Hall Signal Processing Series, C 1983.
    One paper in the book is "Time-Scale Modification of Speech Based on 
    Short-Time Fourier Analysis", M.R. Portnoff, taken from the IEEE 
    transactions on Acoustics, Speech and Signal Processing,Vol 29, No3,
    June 1981.

T.RTitleUserPersonal
Name
DateLines
2700.1SALSA::MOELLEREntropy ThrashWed Aug 21 1991 18:0211
    There is a commercial studio product call the Eventide Harmonizer that
    does time distortion without changing the pitch.  I recall a famous
    conductor going public, disclaiming responsibility for his latest
    record release, because the record company had compressed the symphony
    in time to make it fit in the LP format.
    
    So perhaps contacting Eventide might be an option.  Of course they may
    also consider their interpolation/compression algorhythms to be
    proprietary.
    
    karl
2700.2pitch shiftPIANST::JANZENArthur a grammarWed Aug 21 1991 19:1019
	There are many pitch shifters.  If you can change the speed of
	playback, the pitch shifter will correct the pitch.  However, the
	output is a little fluttery, more so with larger changings of pitch.
	The Yamaha SPX90 does this, so maybe might the Roland DEP 5.
	At much lower prices than the Eventide.  Eventide may or may not
	be considered top of the heap; someone else whose name escapes me
	is.  These expensive units might be higher quality. I'm not sure.
	The conductor probably got a pitch shift by playing the tape faster,
	so the performance was very distorted.  I doubt they used pitch
	shifting, but it's possible I suppose with incredibly high end
	units.
	Radio Shack used to sell a tape recorder with pitch shift
	that I rigged up to accept external line signals; it could only shift
	down (it was for listening to lectures played back fast), and
	was Charge-coupled device-based, and was real hummy and noisey.
	Probaly the best way to synch up music, for which people have a lower
	tolerance of low fidelity (what with the proliferation of
	CDs) is to edit it, either on tape or digitally.
	Tom
2700.3Don't forget lexiconQUIVER::PICKETTDavid - Live free or live in Mass.Wed Aug 21 1991 20:229
    And, of course, there's the digital workstation thingy that Lexicon
    systems won an Academy Award (not Grammy) for, which compresses and
    expands time. The device is used to make music segments fit scenes in
    movies more exactly. 
    
    All that domain transform mumbo jumbo you learned in college really
    works.
    
    dp
2700.4PAULUS::BAUERRichard - ISE L10N Center FrankfurtThu Aug 22 1991 06:507
Hi there !

There's also the AKAI U5 (or is it U4) Phrase Trainer that does both time
compression and pitch correction. And it's a lot cheaper than the other units,
even though the capacity is only for phrases of a few seconds.

	Richard
2700.5MAJTOM::ROBERTThu Aug 22 1991 15:2312
  The DECtalk products from Digital do exactly that, change the speed of
  speech without distorting it.  Now I'm not sure if that was as simple as
  changing the pause duration between spoken words or what.  And of course
  this is voice synthesis, as opposed to real-time modification of incoming
  analog/digital audio.

  But in any case, I would guess the engineers in the DECvoice group would
  be a good place to check.  They must have slews of these sorts of algorithms,
  some of which are probably proprietary, but available for other in-DEC use.

-TR
2700.6Alchemy, for sample waveformsSUBSYS::LYNCHTue Sep 10 1991 18:4513
    If you're interested in doing this with sample waveforms, Alchemy (from
    Passport, formerly from Blank Software) does the job quite well.  As a
    software product, it's quite expensive, but it does things like this
    that other sample editors don't do.  
    
    It does not do this in real time -- it actually recomputes the new
    waveform over a fairly appreciable time interval and then you upload
    the sample back to the synth.  You can, of course, also listen to it
    over the Mac internal audio output.  The computation it does, though,
    is right on the money and has very little in the way of annoying
    distortions or artifacts.
    					-- Mike
    
2700.71% Inspration....OTIGER::R_WHEELERex-Home and GardenMon Sep 23 1991 01:0111
     Sorry I'm late in getting my entry in...  

     I have thought about how to do this for some time, to have correct 
     pitch, but slowed down time.  I've wanted to do this for a band 
     project, and I try to use equipment on hand. My solution (untested) is
     to record the vocals on tape, play through a harmonizer set to, say,
     a fifth above, and sample the output.  Select the root key and play the
     note a fifth below.  The sample should be at the correct pitch, and 
     should be about 1/3 again as long in time as the original.Fine tuning 
     on the pitch control would give more precise control on playback time.  

2700.8somebody has already done itFASDER::AHERBAl is the *first* nameSun Oct 27 1991 01:283
    That is what them 'digital slo-mo" things that are advertised in guitar
    world do isnt it?