IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
file time length (want to see more than 5 decimals)
bomber1978
post Sep 24 2013, 05:43
Post #1





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



Hi guys.
I'm doing some tests on "speed change" in a number of programmes, then creating a mixdown with the original file that cancels out audio to show exactly any difference there may be with the speed changed file once returned to it's original speed compared to the original file...
But to do so I need the speed changed file returned to exactly the original speed, otherwise the mix down doesn't work as it gradually creates more noise as time length progresses because of the time lag that builds up...
Basically all I am looking for is a programme or way to be able to see a file length longer than 5 decimal places (something free). GoldWave athough not a great programme shows file lengths to 5 decimal places, that's the longest I can find, would really appreciate any recommendations on how to see more than 5 decimal places (would like to see as many decimal places as possible!).
PS - I don't need it to even be a good editing programme, as I don't plan on editing with it, all I need is something that can show file time length to many decimal places...
Thanks

This post has been edited by bomber1978: Sep 24 2013, 05:47
Go to the top of the page
+Quote Post
Juha
post Sep 24 2013, 07:49
Post #2





Group: Members
Posts: 368
Joined: 14-February 07
From: EU-FIN
Member No.: 40610



Hmm... can you use sample representation for time (accuracy = 1/samplerate)

Juha

This post has been edited by Juha: Sep 24 2013, 07:56
Go to the top of the page
+Quote Post
bomber1978
post Sep 24 2013, 08:04
Post #3





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



QUOTE (Juha @ Sep 24 2013, 16:49) *
Hmm... can you use sample representation for time (accuracy = 1/samplerate)

Juha


Hi,
Yes I can change milliseconds which shows to 3 decimals, and change it to show samples, I made a 1 second file which showed 44100 samples, and the sample box looks like it is big enough to be able to show exactly the number of samples in my 7min:9sec*****+ file...
This method should work!
Thanks a lot!

This post has been edited by bomber1978: Sep 24 2013, 08:16
Go to the top of the page
+Quote Post
phofman
post Sep 24 2013, 08:14
Post #4





Group: Members
Posts: 261
Joined: 14-February 12
Member No.: 97162



time in secs = nb. of mono/stereo-samples / samplerate in Hz

The most precise figure you can achieve :-)
Go to the top of the page
+Quote Post
bandpass
post Sep 24 2013, 08:42
Post #5





Group: Members
Posts: 321
Joined: 3-August 08
From: UK
Member No.: 56644



Beware of sample and sub-sample delays/padding/truncation that some programmes introduce; the file-length may not relate exactly to the speed in this case.
Go to the top of the page
+Quote Post
bomber1978
post Sep 24 2013, 08:54
Post #6





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



QUOTE (bandpass @ Sep 24 2013, 17:42) *
Beware of sample and sub-sample delays/padding/truncation that some programmes introduce; the file-length may not relate exactly to the speed in this case.


Yeah something isn't exactly right...
I changed an original file (original length 7min:09secs.26667) 8 times with speed change, and have it back at 7min:09.26667.
When I create a mixdown (eg. layer the original file over the file changed speed 8 times), the mixdown starts off very quiet, and gradually gets louder as the mixdown
progresses...
I'm not sure if this is because of time lag because of a difference somewhere after the 5th decimal place (which I can not see)?, or if it is because the speed change is changing the quality - which is the purpose of my test...
The fact that the file starts off very quiet (almost inaudible difference) and gradually gets louder tells me that the difference is most likely to do with time lag, rather than speed change causing an audible change in quality????
Please tell me if I'm wrong, but I think I'm correct??? - If speed change really was creating audible change in quality, the early seconds of the mixdown would not be almost inaudible?, correct?, so the difference I'm hearing has to be because of time lag because I can't get the returned to speed file to match the original after the 5th decimal place???....
Thanks
PS - I'm not sure the sample method suggested is 100% accurate in the programme box I'm using.
As it says 7min:9secs.26667 is 18,930,660 samples, but I used a calculator to calculate 429.26667 X 44100 and got 18,930,660.147 (so .147 difference in sample), that sounds pretty small though so I'm not sure if it's responsible for the time lag I can hear?......

This post has been edited by bomber1978: Sep 24 2013, 09:01
Go to the top of the page
+Quote Post
bandpass
post Sep 24 2013, 10:36
Post #7





Group: Members
Posts: 321
Joined: 3-August 08
From: UK
Member No.: 56644



Another method is to use spectrograms (it doesn't rely on precise time alignment, but equally, doesn't test it). E.g.

CODE
#!/bin/sh

sox -n ref.wav synth -n 10 sin 0:20k gain -1
sox -n in1.wav synth -n 9 sin 0:22.222222222222222k gain -1
sox -n in2.wav synth -n 11.1111111111111111 sin 0:18k gain -1

# You can try other programmes here:
sox in1.wav out1.wav speed 0.9
sox in2.wav out2.wav speed 1.111111111111111111

sox -M ref.wav out1.wav out2.wav all.wav spectrogram -wk -z180 -X 40
display spectrogram.png &



You can check the phase if you want by loading all.wav into audacity etc.
Go to the top of the page
+Quote Post
Juha
post Sep 24 2013, 11:54
Post #8





Group: Members
Posts: 368
Joined: 14-February 07
From: EU-FIN
Member No.: 40610



QUOTE (bomber1978 @ Sep 24 2013, 10:54) *
...
PS - I'm not sure the sample method suggested is 100% accurate in the programme box I'm using.
As it says 7min:9secs.26667 is 18,930,660 samples, but I used a calculator to calculate 429.26667 X 44100 and got 18,930,660.147 (so .147 difference in sample), that sounds pretty small though so I'm not sure if it's responsible for the time lag I can hear?......


0.147 samples means such a short time that, you can't hear that difference (1 sample = 1/44100 seconds (when SR = 44100Hz)).

I suppose that software internally is sample based --> time type representation is calculated and rounded value.


Juha

This post has been edited by Juha: Sep 24 2013, 11:59
Go to the top of the page
+Quote Post
bomber1978
post Sep 24 2013, 16:05
Post #9





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



Thanks for the graph bandpass, I have never looked at using spectographs for this kind of stuff, so I really don't know how to use it to get to the bottom of my questions...
Juha, yes I thought 0.147 samples is too small to have any audible effect between files.

This post has been edited by bomber1978: Sep 24 2013, 16:10
Go to the top of the page
+Quote Post
testyou
post Sep 24 2013, 16:08
Post #10





Group: Members
Posts: 91
Joined: 24-September 10
Member No.: 84113



Which program are you using bomber?

Program - Wikipedia
Go to the top of the page
+Quote Post
bomber1978
post Sep 24 2013, 16:11
Post #11





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



QUOTE (testyou @ Sep 25 2013, 01:08) *
Which program are you using bomber?

Program - Wikipedia


Hi,
For these particular files, Audacity 2.0.3.
This has kind of become another thread on the board, where I have posted the actual files I'm refering to with more detail, sorry it started as a different question, but it's pretty much ended up on the same subject, I think it probably explains things better here:
http://www.hydrogenaudio.org/forums/index....howtopic=102773

This post has been edited by bomber1978: Sep 24 2013, 16:18
Go to the top of the page
+Quote Post
bandpass
post Sep 24 2013, 17:09
Post #12





Group: Members
Posts: 321
Joined: 3-August 08
From: UK
Member No.: 56644



QUOTE (bomber1978 @ Sep 24 2013, 16:05) *
I have never looked at using spectographs for this kind of stuff, so I really don't know how to use it to get to the bottom of my questions...

The graph shows the amount of aliasing/imaging distortion introduced by the speed-change, and assuming this is the concern then unless the level is much higher than shown (perhaps red-ish on the above scale) then it's probably not worth even trying to ABX.
Go to the top of the page
+Quote Post
bomber1978
post Sep 24 2013, 17:37
Post #13





Group: Members
Posts: 27
Joined: 24-September 13
Member No.: 110284



QUOTE (bandpass @ Sep 25 2013, 02:09) *
QUOTE (bomber1978 @ Sep 24 2013, 16:05) *
I have never looked at using spectographs for this kind of stuff, so I really don't know how to use it to get to the bottom of my questions...

The graph shows the amount of aliasing/imaging distortion introduced by the speed-change, and assuming this is the concern then unless the level is much higher than shown (perhaps red-ish on the above scale) then it's probably not worth even trying to ABX.


Sorry I just don't understand the graph!, I can see 0 to -180 decibels on the right side but I'm just not sure how to apply it to a file!, I don't even know what you mean by "You can check the phase if you want by loading all.wav into audacity etc.".
I'm presuming this is for the "speed changed file", and not the "mixdown file", I'll try and work it out!?!?!?

This post has been edited by bomber1978: Sep 24 2013, 17:41
Go to the top of the page
+Quote Post
2Bdecided
post Sep 24 2013, 17:51
Post #14


ReplayGain developer


Group: Developer
Posts: 4945
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



Speed change without preserving pitch is just resampling. Those graphs are one good way of looking at the mistakes a resampler might be making: the diamond patterns are unwanted aliasing.

See threads on resampling for more info.

Cheers,
David.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 20th April 2014 - 18:09