Help - Search - Members - Calendar
Full Version: how do I remove digital silence from a stereo .wav file?
Hydrogenaudio Forums > Hydrogenaudio Forum > General Audio
dissociative
I need to remove leading and trailing digital silence without modifying the pcm data of a wav file in anyway, The only tool that I have is audacity but I'm unsure if it can do it losslessly as it converts the pcm samples from 16-bit integer to 32-bit float by default, This .wav contains the background music of a game, it was produced/recorded with a sound Plugin of the playstation emulator ePSXe, but I recorded it just some seconds before the song started to play and the audio ends some seconds after the song finalizes playing.
uart
QUOTE(dissociative @ Feb 3 2008, 17:36) *

I need to remove leading and trailing digital silence without modifying the pcm data of a wav file in anyway, The only tool that I have is audacity but I'm unsure if it can do it losslessly as it converts the pcm samples from 16-bit integer to 32-bit float by default, This .wav contains the background music of a game, it was produced/recorded with a sound Plugin of the playstation emulator ePSXe, but I recorded it just some seconds before the song started to play and the audio ends some seconds after the song finalizes playing.


Don't worry, the entire process of going from 16 bit integer to 32bit float and then back to 16 bit integer is completely lossless. The mantissa of a 32bit float is longer than 16 bits and that's all you need to ensure there is no truncation.
fj4
Have you looked into PSXMC? Thank Zophar's Domain for hosting it.
It can convert PlayStation .xa sound files to wav, right from the disc.

Otherwise, yeah, Audacity is useful as always!
SpasV
smile.gif The mantissa of a 32 bits float is 24 bits, the exponent is 8 bits.
Really, 24 bits mantissa is far enough for a 16 bits integer. (Both use one sign bit.)
uart
Interesting that although there is absolutely no technical reason why 16 bit integer to 32bit float and back to 16 bit integer shouldn't be bit perfect I just opened a wav file (normal 16 bit) and then exported to wav and it wasn't the case. sad.gif

I had errors of +/- 8 LSB's between the original and saved files. WTF?

BTW I've got audacity ver 1.23 here. The latest stable is 1.26 so maybe a later version can do better. Can someone else test this as well, it really shouldn't happen!
Trancer
QUOTE(uart @ Feb 4 2008, 12:12) *

Interesting that although there is absolutely no technical reason why 16 bit integer to 32bit float and back to 16 bit integer shouldn't be bit perfect


This i where you go wrong.
dissociative
QUOTE(fj4 @ Feb 4 2008, 02:13) *

Have you looked into PSXMC? Thank Zophar's Domain for hosting it.
It can convert PlayStation .xa sound files to wav, right from the disc.

Otherwise, yeah, Audacity is useful as always!


PSXMC will not work because the music of the game is produced from some kind of sound Sequencer

well I was able to remove the leading digital silence of the wav file using Exact Audio Copy 0.99pb3 in "Tools > Process WAV..." but the song seems to end with a non-digital silence pattern that I cannot hear and lasts forever no matter how many time I spend recording in the end.
lvqcl
QUOTE(uart @ Feb 4 2008, 14:12) *

Interesting that although there is absolutely no technical reason why 16 bit integer to 32bit float and back to 16 bit integer shouldn't be bit perfect I just opened a wav file (normal 16 bit) and then exported to wav and it wasn't the case. sad.gif

I had errors of +/- 8 LSB's between the original and saved files. WTF?


Dithering, I suppose
uart
QUOTE(lvqcl @ Feb 4 2008, 08:31) *

QUOTE(uart @ Feb 4 2008, 14:12) *

Interesting that although there is absolutely no technical reason why 16 bit integer to 32bit float and back to 16 bit integer shouldn't be bit perfect I just opened a wav file (normal 16 bit) and then exported to wav and it wasn't the case. sad.gif

I had errors of +/- 8 LSB's between the original and saved files. WTF?


Dithering, I suppose

Yes dithering is turned on by default in the preferecnes, after turning it off it's bit perfect as expected.
PiezoTransducer
Have you tried WavTrim?
http://www.mptrim.com/WavTrim.html
mazz90
You could try to FlexiMusic Wave Editor, it have used it to remove and delete the silence, i got good output.
Kees de Visser
QUOTE(Trancer @ Feb 4 2008, 16:59) *

QUOTE(uart @ Feb 4 2008, 12:12) *

Interesting that although there is absolutely no technical reason why 16 bit integer to 32bit float and back to 16 bit integer shouldn't be bit perfect
This i where you go wrong.
Could you please explain why ?
IMO the conversion should be transparent for 16-bit signals unless some processing is applied that modifies the 16-bit signal.
Any 32-bit artifacts that remain well below 16-bit LSB level will disappear when truncating back to 16-bit.
SebastianG
QUOTE(Kees de Visser @ Feb 21 2008, 09:26) *

Could you please explain why ?

answer:
QUOTE(uart @ Feb 6 2008, 00:50) *

QUOTE(lvqcl @ Feb 4 2008, 08:31) *

Dithering, I suppose

Yes dithering is turned on by default in the preferecnes, after turning it off it's bit perfect as expected.

This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.