Become a Site Supporter and Never see Ads again!

Author Topic: To shrink or stretch?  (Read 4493 times)

0 Members and 1 Guest are viewing this topic.

Offline danny3

  • Trade Count: (10)
  • Taperssection Regular
  • **
  • Posts: 212
  • Gender: Male
To shrink or stretch?
« on: July 20, 2014, 02:30:23 PM »
It took me a while but I now have applied advice found on TS to mix files from two different recorders with Audacity. The method I use is to line up the files at the beginning, find a reference point at the end, divide longer sample by the lower sample, apply percentage in the Change Speed effect.

Currently working on a recording from last night, where I had 4 channels (stage mics / aud mics) going into a DR-680. I was sent the sbd by a fellow taper, and have entered that into my project. The sbd file is shorter, and using the above method I selected the two stereo files from the 680 and applied .00314 % with the Change Speed tool. The set was @ 58 minutes. (Set 2 is 1:42)

This worked fine, but I question if it would be better to change (stretch?) the sbd to match the other files? I don’t know how to do this yet, but will seek out how to do so. It was a pretty small adjustment, and I don’t know how much the pitch/tempo could be adversely affected. Any thoughts would be greatly appreciated. Thanks.

Offline dmonkey

  • Trade Count: (7)
  • Taperssection All-Star
  • ****
  • Posts: 1056
  • Gender: Male
Re: To shrink or stretch?
« Reply #1 on: July 20, 2014, 02:40:23 PM »
I've never noticed a difference and I've done both several times. My guess is that it doesn't matter in a practical sense (perhaps there's some theoretical argument to be made, though).
MK4's, KM140's or MC930's >  Tinybox or Aerco MP-2 > R-09, M-10 or R-44 (Oade CM)

Offline 2manyrocks

  • Trade Count: (12)
  • Taperssection All-Star
  • ****
  • Posts: 1645
Re: To shrink or stretch?
« Reply #2 on: July 20, 2014, 03:41:05 PM »
Another option is to split one of the audio files during a silent moment  as needed and then align each segment without shrinking or stretching anything.  Just listen to the recording and make your splits/adjustments between songs as needed. 


Offline danny3

  • Trade Count: (10)
  • Taperssection Regular
  • **
  • Posts: 212
  • Gender: Male
Re: To shrink or stretch?
« Reply #3 on: July 20, 2014, 04:04:15 PM »
Another option is to split one of the audio files during a silent moment  as needed and then align each segment without shrinking or stretching anything.  Just listen to the recording and make your splits/adjustments between songs as needed.

Aligning each song separately is the approach I used until I got my head around the speed change method, which is actually very easy, except for times when it is near impossible to find a good point of reference.


Offline Sloan Simpson

  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 3480
  • Gender: Male
    • Southern Shelter
Re: To shrink or stretch?
« Reply #4 on: July 20, 2014, 05:03:06 PM »
I would think with modern software shrinking or stretching would maintain accuracy much better than splicing and realigning.

To the original post I'd go with the "if it sounds good it probably is good" rule of thumb. Either stretching or shrinking should work just as well.
Neumann KM-184> Tascam DR-680

runonce

  • Guest
  • Trade Count: (0)
Re: To shrink or stretch?
« Reply #5 on: July 20, 2014, 05:35:45 PM »
I'd alter the source I'd be using the least of in the mix....

The chop and align method is flawed because the sources are slightly out of tune...avoid.

Offline page

  • Trade Count: (25)
  • Needs to get out more...
  • *****
  • Posts: 8392
  • Gender: Male
  • #TeamRetired
Re: To shrink or stretch?
« Reply #6 on: July 20, 2014, 05:47:43 PM »
I'd alter the source I'd be using the least of in the mix....

The chop and align method is flawed

Agreed. When I was doing this, I tweaked the source I was going to use the least amount of. And yes, the chop method sucks because all it takes is 4 or 5 minutes to hear even small changes in the soundstage where stuff smears due to time differences.
"This is a common practice we have on the bus; debating facts that we could easily find through printed material. It's like, how far is it today? I think it's four hours, and someone else comes in at 11 hours, and well, then we'll... just... talk about it..." - Jeb Puryear

"Nostalgia ain't what it used to be." - Jim Williams

Offline 2manyrocks

  • Trade Count: (12)
  • Taperssection All-Star
  • ****
  • Posts: 1645
Re: To shrink or stretch?
« Reply #7 on: July 20, 2014, 08:53:51 PM »
I see what you are both saying. 

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (13)
  • Needs to get out more...
  • *****
  • Posts: 12131
  • Gender: Male
Re: To shrink or stretch?
« Reply #8 on: July 20, 2014, 09:54:47 PM »
I doubt it makes much audible difference whether you stretch or shrink, if only because the degree of change is rather minimal given the overall length of the files.  Like the other fellows mentioned, probably best practice to modify the least significant source in the mix.

When I've done this in the past with two sources used more or less equally, I opted to shrink the longer file with the reasoning that condensing what information is already there (going from 'more' to 'less') may arguably be somewhat less artifact prone than stretching that existing limited information out (going from 'less' to 'more'). Sort of like a sample rate reduction recalculation or bit-depth reduction throws out some of the existing information, and a sample rate or bit-depth increase doesn't add any additional information beyond what was in the original file.

If you want to find out if one is actually better, at least for more significant changes, you could do a test and adjust two copies a clip by a large percentage both ways, maybe making the same clip twice and half as long.  Listen for which one withstands the time-manipulation algorithm with the least objectionable distortions.
« Last Edit: July 20, 2014, 09:57:13 PM by Gutbucket »
volition > vibrations > voltages > numeric values | numeric values > voltages > vibrations> virtual teleportation time-machine experience
made easy- >>Improved PAS table<< | made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<<

Offline montanamon

  • Trade Count: (0)
  • Taperssection Newbie
  • *
  • Posts: 11
  • NeumannKM-184's (matched pair), EdirolR-4Pro (Oade
Re: To shrink or stretch?
« Reply #9 on: July 24, 2014, 10:33:52 AM »
I also doubt the shrink vs stretch has much audible difference when were making minimal adjustments over a longer length of time. I've generally used the SBD (the larger % of the mix) as the "baseline" for time and adjusted the AUD file to match using a similar adjustment in Audacity. The matmatical formula you mentioned also mentioned in past forums here on TS regarding post show matrix's i haven't found to be a perfect art. Often after the math was done i was finding small differences when i thought the math was solid. I have since that point been literally eyeing the final result ensuring they match exactly with phasing being the digitial artifact that i have found if you don't.

Has anyone used Sony Soundforge (version 10 or 11) for time stretching or matching up SBD vs AUD recorded on different recorders ? Or can anyone else recommend a better program that is spot on for stretching or shrinking as Audacity certainly works however, i just don't find very user friendly in this matter...

Any possible changes with pitch or tempo is really the question i have - is there anyone that could chime in on this topic in relation to a stretch or shrink ? I know Charlie Miller recordings (of Grateful Dead fame) are often checked for proper pitch but i don't know enough about the topic and how it could relate here.

Finally my own two cents (from personal experience) is that having chopped files to individually lay over a "baseline" file for a matrix still would require subtle shrink or stretch issues with the timeline and i have often had a few phasing issues with that. I also do not recommend that approach...

Offline bombdiggity

  • Trade Count: (11)
  • Needs to get out more...
  • *****
  • Posts: 2212
Re: To shrink or stretch?
« Reply #10 on: July 24, 2014, 12:10:01 PM »
^ I don't think which of shrink or stretch is employed will matter (assuming a digital realm since the net change there should be infinitesimal).  Of course by doing either of those one is changing the pitch (very subtly I'd presume).  Likely only a professional musician could say which of the sources is objectively at exactly the right speed and thus pitch (if either of them is actually spot on).  If dealing with fractions of cents the correct absolute pitch may not even be identifiable.  The rule of thumb from my golden eared music degreed friend (who is truly Mr. Wonder Pitch) is if a source is less than 10 cents off he doesn't fool with it (it's a rare ear that can hear that small a difference in absolute terms, though clearly if you're aligning sources that would be a huge variation that would be impossible to reconcile without the requisite change).  The possibility of any digital sources being anywhere near that far different from each other seems to be zero (unless DAT tape is played back at the wrong sample rate or malfunctioning). 

I would actually recommend against doing any pitch adjustment in Audacity.  My friend says he felt that program has unacceptable impacts on the quality when doing anything more than very basic processing (levels, tracking).  Pitch adjustment is a high end function that should be done in high end software.  Which seems a matter of choice or access, though people have their favorites, there are differences and workflows can be quite different. 

The "chopping" approach is very crude and will only result in memorializing any drift between the two sources (rather than getting it right it basically only ensures that the initial match points may match but nothing else will). 
Gear:
Audio:
Schoeps MK4V
Nak CM-100/CM-300 w/ CP-1's or CP-4's
SP-CMC-25
>
Oade C mod R-44  OR
Tinybox > Sony PCM-M10 (formerly Roland R-05) 
Video: Varied, with various outboard mics depending on the situation

Offline 2manyrocks

  • Trade Count: (12)
  • Taperssection All-Star
  • ****
  • Posts: 1645
Re: To shrink or stretch?
« Reply #11 on: July 24, 2014, 12:54:50 PM »
Which high end software does your friend find acceptable? 

Can you shrink or stretch in Reaper?  Is it any better than Audacity for this?

For all the trouble in aligning between different recorders, I'm sort of thinking of selling my Dr-2d and buying another M10.  Wondering if two M10s will record closely enough to avoid any alignment issues in recordings under an hour and a half?

Offline bombdiggity

  • Trade Count: (11)
  • Needs to get out more...
  • *****
  • Posts: 2212
Re: To shrink or stretch?
« Reply #12 on: July 24, 2014, 01:31:39 PM »
Which high end software does your friend find acceptable? 

Can you shrink or stretch in Reaper?  Is it any better than Audacity for this?

For all the trouble in aligning between different recorders, I'm sort of thinking of selling my Dr-2d and buying another M10.  Wondering if two M10s will record closely enough to avoid any alignment issues in recordings under an hour and a half?

My friend uses Audition but some complain about its work flow (I'm not a huge fan of the work flow but have come to learn it OK).  Reaper is well liked around here and I know one of the multichannel recordists relies on it.  Sound Forge, Wave Lab...  Basically any of the better retail software is going to be better (and probably easier to do this in).

I'd think similar recorders would align better but they may not be exact either.  Of course if you use the same recorders all the time you should be able to identify the exact offset between them.  I'd be shocked if the difference once you align the starting point was not always the same setting in absolute terms.  Of course if you're correcting by applying a % to varying length sets then you don't know what the absolute offset is and are likely going about it wrong (by applying a variable adjustment methodology to a fixed differential).  If Audition only lets you do % then it is a sign it is lacking in that area. 

Pitch correction should always be done in absolute terms (in cents/semi-tones) not by %.  Anyone who adjusts pitch and says they corrected it by such and such % isn't doing it right.  This is time alignment rather than pitch correction but I think the principle is the same.  The recorders should vary by an absolute amount (a certain number of milliseconds or samples per hour or whatever increment you choose).  Once you conclusively identify that offset you should be able to just apply that rate to every instance. 

The more obvious solution if you run multiple sources is to get a multichannel recorder, though the practical geography of that may not work if you're running mics and board since the sweet spot is rarely near the board. 
« Last Edit: July 24, 2014, 03:54:29 PM by bombdiggity »
Gear:
Audio:
Schoeps MK4V
Nak CM-100/CM-300 w/ CP-1's or CP-4's
SP-CMC-25
>
Oade C mod R-44  OR
Tinybox > Sony PCM-M10 (formerly Roland R-05) 
Video: Varied, with various outboard mics depending on the situation

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (13)
  • Needs to get out more...
  • *****
  • Posts: 12131
  • Gender: Male
Re: To shrink or stretch?
« Reply #13 on: July 24, 2014, 03:34:30 PM »
I'm doing it in Samplitude, using the most appropriate setting in their advanced algorithm which maintains pitch while adjusting the timebase, and that partly figured into my decision to shrink the longer instead of stretch the shorter.  Neither source is particularly dominant my case because these are Left/Right/Center/Surround recordings, L/R on one recorder, C/S on the other.  Sometimes these are mixed to 2-channel, though primarily synced and left as discrete channels which still need to be in phase alignment (at least the L/C/R channels do, the S channel is more forgiving by a few ms long as it doesn't lead).

It's pretty straightforward to do in Samplitude at the object level, using that algorithm.
volition > vibrations > voltages > numeric values | numeric values > voltages > vibrations> virtual teleportation time-machine experience
made easy- >>Improved PAS table<< | made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<<

Offline Fried Chicken Boy

  • Site Supporter
  • Trade Count: (8)
  • Needs to get out more...
  • *
  • Posts: 2932
Re: To shrink or stretch?
« Reply #14 on: July 24, 2014, 04:15:01 PM »
Good advice and suggestions in this thread.  Going to try putting together a matrix of two different sources and this should come in handy.  +T

 

RSS | Mobile
Page created in 0.247 seconds with 42 queries.
© 2002-2018 Taperssection.com
Powered by SMF