Region gain boosted when region is time-stretched

I’m not sure if this is a bug, as I can’t seem to find other people having this problem.

When I use the time stretch tool to stretch a region, the gain on the region is boosted.

I don’t know much about automation (I haven’t touched the automation envelopes), but I can see that the track’s gain automation envelope is also altered after the time stretch.

Am I doing something wrong? Have I altered a setting in ardour that needs to be reset?

Any help is appreciated.

Thanks in advance,


Devin

When I say gain envelope, I’m referring to the automation area for the gain. I’m probably using the wrong terminology, and it’s possible (I don’t know how it works) that the change in the gain of the audio data is causing the change in the automation data.

I will export the region and do the analysis a little later today.

If and when I log on to IRC, my handle will begin with “SurfacePatterns”.

Thank you for all your help. I will get the analysis data to you ASAP.


Devin

could you arrange some screenshots to show the alteration?

btw, my usual reminder: forums are horrible places to try to solve problems with software. IRC (see http://ardour.org/support) is much better in every way: faster, more immediate, more flexible etc. etc.

Thank you for your quick reply.

Should I repost this topic elsewhere (the users’ mailing list, perhaps), or should I continue to use the forum for this instance?

I’ll get you some screenshots ASAP.


Devin

Oh. Duh. IRC. Right.


Devin

Here are some screenshots. I’ll go to the IRC channel if I need to as well. Watch the region just before the split.

Before:

During:

After:


Devin

OK, it appears that when you say the “gain envelope” has changed, you are not referring to the region gain envelope, but the actual audio data itself. correct? if you are on IRC, it would be useful to know your nick/handle there.

you might want to try exporting the region, installing rubberband (the library+tools we use for stretching) and using the rubberband utility to do the stretch outside ardour, then use sndfile-info to compare the peak amplitudes. if it changes dramatically, this would be a rubberband issue.

the problem is that i don’t actually see any gain automation data in those screenshots. the display in the “gain” track is just a “shadow” of the waveform data in the main track. it has gotten bigger in the same way that the waveform in the main track has gotten bigger.

see you on IRC …

I don’t yet understand automation. The “shadow” was what I was referring to.

I went onto IRC, and Hans told me that the main developers were all asleep.

I’ve done the stretching in and out of ardour.

Here is the rubberband command I used to stretch the sound file:

[devin@localhost kill-for-you]$ rubberband --time 1.4736328125 before-stretch.wav after-rubberband-stretch.wav
Using crispness level: 4 (Crisp monophonic instrumental)
Pass 1: Studying...
Calculating profile...
Pass 2: Processing...

in: 22528, out: 33198, ratio: 1.47363, ideal output: 33198, error: 0
elapsed time: 0.110015 sec, in frames/sec: 204772, out frames/sec: 301759

Here’s the output from sndfile-info of the file before stretching, the file after being stretched in ardour, and the file after being stretched using rubberband:

[devin@localhost kill-for-you]$ sndfile-info before-stretch.wav after-ardour-stretch.wav after-rubberband-stretch.wav

Version : libsndfile-1.0.17

========================================
File : before-stretch.wav
Length : 67628
RIFF : 67620
WAVE
fmt  : 16
  Format        : 0x1 => WAVE_FORMAT_PCM
  Channels      : 1
  Sample Rate   : 44100
  Block Align   : 3
  Bit Width     : 24
  Bytes/sec     : 132300
data : 67584
End

----------------------------------------
Sample Rate : 44100
Frames      : 22528
Channels    : 1
Format      : 0x00010003
Sections    : 1
Seekable    : TRUE
Duration    : 00:00:00.510
Signal Max  : 394392 (-26.56 dB)

========================================
File : after-ardour-stretch.wav
Length : 99638
RIFF : 99630
WAVE
fmt  : 16
  Format        : 0x1 => WAVE_FORMAT_PCM
  Channels      : 1
  Sample Rate   : 44100
  Block Align   : 3
  Bit Width     : 24
  Bytes/sec     : 132300
data : 99594
End

----------------------------------------
Sample Rate : 44100
Frames      : 33198
Channels    : 1
Format      : 0x00010003
Sections    : 1
Seekable    : TRUE
Duration    : 00:00:00.752
Signal Max  : 684114 (-21.77 dB)

========================================
File : after-rubberband-stretch.wav
Length : 99638
RIFF : 99630
WAVE
fmt  : 16
  Format        : 0x1 => WAVE_FORMAT_PCM
  Channels      : 1
  Sample Rate   : 44100
  Block Align   : 3
  Bit Width     : 24
  Bytes/sec     : 132300
data : 99594
End

----------------------------------------
Sample Rate : 44100
Frames      : 33198
Channels    : 1
Format      : 0x00010003
Sections    : 1
Seekable    : TRUE
Duration    : 00:00:00.752
Signal Max  : 359911 (-27.35 dB)

As you can see, the peak amplitude of the ardour result is significantly higher than the peak amplitude of the rubberband result.

I’ll try to log in to IRC at some point tomorrow.

Thank you for your help. :slight_smile:


Devin