IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
How to convert WAV 32 bit float to WV 32 bit float
David the Chanse...
post Oct 9 2013, 21:58
Post #1





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



I have audio files in WAV (32 bit float, 96kHz) format, and I wish to compress them (losslessly) to save space. FLAC doesn't support 32 bit float if I am not mistaken, so I am hoping to convert these files to WV instead. How can I go about this? Is it possible/recommended to use Audacity? I have downloaded WavPack Win32 and the Winamp Plugin from wavpack's website, but I am unsure how to go about converting the WAV files retaining the 32 bit float bit rate. Any help would be greatly appreciated.
Go to the top of the page
+Quote Post
saratoga
post Oct 9 2013, 22:04
Post #2





Group: Members
Posts: 4715
Joined: 2-September 02
Member No.: 3264



I would use a tool that can do batch conversions, like foobar2000 or dbpoweramp, then set the appropriate settings (lossless mode, 32 bit float) in the gui.
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 9 2013, 22:50
Post #3





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



QUOTE (saratoga @ Oct 9 2013, 22:04) *
I would use a tool that can do batch conversions, like foobar2000 or dbpoweramp, then set the appropriate settings (lossless mode, 32 bit float) in the gui.


I downloaded the latest version of foobar2000 (v1.2.9) and converted 1 of the WAV files to WV 32 bit (it doesn't specify that it's float). I then played it in Winamp with the plugin but the 1:55 file played extremely quickly, lasting just 10-15 seconds. It didn't sound like a sped up version of the original file either, it was just noise. Was the conversion 32 bit float? And why did the file play like this?

Edit: in the configurations for WavPack Decoder 2.8a in Input Plugins in Winamp, checking the "always output 16-bit audio" box makes the playback work. However, I would ideally prefer 24 bit playback. The WV file plays fine in foobar2000.

This post has been edited by David the Chansey: Oct 9 2013, 22:57
Go to the top of the page
+Quote Post
saratoga
post Oct 9 2013, 22:59
Post #4





Group: Members
Posts: 4715
Joined: 2-September 02
Member No.: 3264



Edit: saw that you figured it out.

Do you actually need winamp at all? Generally one uses floating point for editing. If you're done editing and want to actually listen to the files I would convert them to something more compatible.

This post has been edited by saratoga: Oct 9 2013, 23:01
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 9 2013, 23:12
Post #5





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



QUOTE (saratoga @ Oct 9 2013, 22:59) *
Edit: saw that you figured it out.

Do you actually need winamp at all? Generally one uses floating point for editing. If you're done editing and want to actually listen to the files I would convert them to something more compatible.


I'd prefer to keep them at 32 bit float for archiving, but also be able to play them with no need for more conversion. I don't need Winamp, but I do like it. I'll see if I like foobar, and stick to it if I do.

Also, is the WV file actually 32 bit float? The option for bit depth when converting only said 32 bit.
Go to the top of the page
+Quote Post
eahm
post Oct 10 2013, 00:46
Post #6





Group: Members
Posts: 882
Joined: 11-February 12
Member No.: 97076



QUOTE (David the Chansey @ Oct 9 2013, 15:12) *
Also, is the WV file actually 32 bit float? The option for bit depth when converting only said 32 bit.

http://www.wavpack.com/ read the Features section.

4.70 stable is coming soon.

This post has been edited by eahm: Oct 10 2013, 00:46
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 10 2013, 00:56
Post #7





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



QUOTE (eahm @ Oct 10 2013, 00:46) *
QUOTE (David the Chansey @ Oct 9 2013, 15:12) *
Also, is the WV file actually 32 bit float? The option for bit depth when converting only said 32 bit.

http://www.wavpack.com/ read the Features section.

4.70 stable is coming soon.


Which part did you want me to read in particular? I'm trying to convert a 32 bit float WAV file to WV, retaining the 32 bit float bit rate.
Go to the top of the page
+Quote Post
greynol
post Oct 10 2013, 01:01
Post #8





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



If you don't mind my asking, what is the content that requires such precision?


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 10 2013, 01:10
Post #9





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



QUOTE (greynol @ Oct 10 2013, 01:01) *
If you don't mind my asking, what is the content that requires such precision?

They are tunes that I have made and exported as 32 bit float wave files. I want to keep them at this bit rate just for the peace of mind that it's the highest bit rate possible - to be honest - in case I decide I want to use them in the production of future music. I also want to be able to play them without having to make additional converted copies. I'm the type of person who is anal about high quality, even if it is inaudible (lol).

The way I see it is if I can keep it at that bit rate and play them back ok, then I don't mind the cost of extra storage space.
Go to the top of the page
+Quote Post
greynol
post Oct 10 2013, 01:16
Post #10





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



So long as you know the stuff below ~19-20 bits will never be reproduced with any consistency upon playback, which shouldn't matter because it is, in all likelihood, only noise anyway.


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 10 2013, 01:34
Post #11





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



Noise is all part of the music. wink.gif I see that the playback quality will not be improved with 32f since my audio interface plays 24 bit audio anyway. I just want to keep them as high quality as possible (even if the difference is ridiculously minuscule) and be able to play them back.

I have discovered that foobar's bit rate setting is indeed 32 bit float, not 32 bit integer. Also, foobar offers the "additional processing" option for further compression whereas dBpoweramp does not (at least in the trial version). I'll use foobar for some time and I'll probably end up liking it over winamp, like many others do. Thanks for the help.
Go to the top of the page
+Quote Post
db1989
post Oct 10 2013, 07:49
Post #12





Group: Super Moderator
Posts: 5141
Joined: 23-June 06
Member No.: 32180



QUOTE (David the Chansey @ Oct 10 2013, 01:34) *
Noise is all part of the music. wink.gif
Not when (A) it is so far down in volume as to be inaudible and (B) your interface almost certainly just made the noise up arbitrarily with no relation to the music and then dumped it on top.

QUOTE
I see that the playback quality will not be improved with 32f since my audio interface plays 24 bit audio anyway.
Even if your interface could play 32-bit audio, it would make no difference in audible terms (unless its implementation of 24-bit audio was hopelessly broken).

QUOTE
I just want to keep them as high quality as possible (even if the difference is ridiculously minuscule) and be able to play them back.
I can understand a compulsion not to downsample the files if you can avoid it. Saying it will ever matter audibly is a different matter.
Go to the top of the page
+Quote Post
phofman
post Oct 10 2013, 08:42
Post #13





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



Checking whether the wav file is float32 format:
CODE
soxi float32.wav

Input File     : 'float32.wav'
Channels       : 2
Sample Rate    : 44100
Precision      : 24-bit
Duration       : 00:01:34.93 = 4186560 samples = 7120 CDDA sectors
File Size      : 33.5M
Bit Rate       : 2.82M
Sample Encoding: 32-bit Floating Point PCM


Econding to wavpack using http://www.wavpack.com/downloads.html
CODE
wavpack float32.wav

WAVPACK  Hybrid Lossless Audio Compressor  Linux Version 4.60.1
Copyright (c) 1998 - 2009 Conifer Software.  All Rights Reserved.

created float32.wv in 2.67 secs (lossless, 79.32%)


Checking resultant float32.wv format:
CODE
soxi float32.wv

Input File     : 'float32.wv'
Channels       : 2
Sample Rate    : 44100
Precision      : 24-bit
Duration       : 00:01:34.93 = 4186560 samples = 7120 CDDA sectors
File Size      : 6.93M
Bit Rate       : 584k
Sample Encoding: 32-bit Floating Point WavPack
Go to the top of the page
+Quote Post
2Bdecided
post Oct 10 2013, 12:09
Post #14


ReplayGain developer


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



QUOTE (David the Chansey @ Oct 10 2013, 01:10) *
I also want to be able to play them without having to make additional converted copies.
Support for 32-bit float wav files, never mind losslessly compressed versions of them, is pretty patchy. Worse, when some software tries to play them, you just get full scale digital noise. Some software will play them, but downconvert them, and not necessarily in an optimal way (though audible problems shouldn't exist).

By all means keep them, and by all means losslessly compress them - but maybe make 96/24 and/or 44.1/16 copies for listening to.

Cheers,
David.
Go to the top of the page
+Quote Post
Nick.C
post Oct 10 2013, 12:56
Post #15


lossyWAV Developer


Group: Developer
Posts: 1772
Joined: 11-April 07
From: Wherever here is
Member No.: 42400



QUOTE (2Bdecided @ Oct 10 2013, 12:09) *
By all means keep them, and by all means losslessly compress them - but maybe make 96/24 and/or 44.1/16 copies for listening to.


.... or Float16 WAV versions to play in foobar2000 or RockBox.

I would expect that foobar2000 v1.3.0 will play 16-bit (Half Precision) or 24-bit float (truncated / rounded 32-bit float) encoded in WavPack using the --store-floats-as-ints parameter introduced in 4.70.0 beta.


--------------------
lossyWAV -q X -a 4 --feedback 4| FLAC -8 ~= 320kbps
Go to the top of the page
+Quote Post
pdq
post Oct 10 2013, 13:37
Post #16





Group: Members
Posts: 3304
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



QUOTE (David the Chansey @ Oct 9 2013, 20:10) *
I want to keep them at this bit rate just for the peace of mind that it's the highest bit rate possible - to be honest - in case I decide I want to use them in the production of future music.

You are mistaken. Not that it matters, but 32 bit float has only 24 bits of mantissa, so it is inferior to 32 bit integer. The only advantage of 32 bit float is that it has a much higher dynamic range, but for audio data this is quite useless.
Go to the top of the page
+Quote Post
David the Chanse...
post Oct 10 2013, 17:36
Post #17





Group: Members
Posts: 7
Joined: 9-October 13
Member No.: 110810



Sorry, I forgot to mention that I couldn't export the files in 32 bit integer, only 32 bit float. I'll convert them to 24bit, 96kHz FLAC for playback, as I have had no problems with FLAC playback so far.

One last question: is 24 bit audio actually going to my audio interface? I.e. does foobar, or any other music player, convert it to 16 bit?

Thanks for your help, and sorry for my stubbornness of wanting to keep 32 bit float.
Go to the top of the page
+Quote Post
saratoga
post Oct 10 2013, 19:16
Post #18





Group: Members
Posts: 4715
Joined: 2-September 02
Member No.: 3264



QUOTE (David the Chansey @ Oct 10 2013, 12:36) *
One last question: is 24 bit audio actually going to my audio interface? I.e. does foobar, or any other music player, convert it to 16 bit?


Depends on what hardware you have. Assuming its 24 bit capable, its probably using 24 bit. That said, almost everything these days will take 24 bit audio, although not all that many devices actually produce more than 16 effective bits.
Go to the top of the page
+Quote Post
2Bdecided
post Oct 11 2013, 10:00
Post #19


ReplayGain developer


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



QUOTE (pdq @ Oct 10 2013, 13:37) *
You are mistaken. Not that it matters, but 32 bit float has only 24 bits of mantissa, so it is inferior to 32 bit integer. The only advantage of 32 bit float is that it has a much higher dynamic range, but for audio data this is quite useless.
You don't have to stay below "digital full scale" because it no longer is full scale. This is very useful when processing: no need to even think about headroom, just do whatever you like. Whether the result is far too loud or far too quiet, you always have 24-bits of precision relative to the peak, and you never get clipping.

With conventional audio processing, there are usually easy enough ways to keep the results within range anyway, but with experimental stuff the ability to go above digital full scale without issue can be very helpful.

You have to bring it back into range to listen to it. wink.gif But I'd argue that it's still audio.

Cheers,
David.
Go to the top of the page
+Quote Post
Nystagmus
post Nov 9 2013, 20:34
Post #20





Group: Members
Posts: 22
Joined: 13-October 13
Member No.: 110926



I am a musician and I use 32-bit float WAV's regularly. It's worth adding that in MultitrackStudio Pro Plus, at least in the old versions, freezing VSTi's could cause clipping UNLESS 32-bit float was used. Also, using CoolEdit2000 or similar, if the original file was a 32-bit float, the otherwise clipped audio could be normalised DOWN to 0 dBfs or just below.

It's for this reason that I've been using 32-bit float for many years. It's a good habit to avoid clipping during composing, but with some extravagant VST's and VSTi's chained together, it starts to get a bit more difficult. So I really do like having the peace of mind knowing that if I accidentally clip somthing I can just fix it with a simple normalise or appropriate gain command.

And like some other folks, I like knowing that my tunes are rendered to their fullest bit-resolution quality possible using 32-bit float. Naturally, wanting to find a lossless compression and playback utility set, I came across foobar2000 as one of the only players that could do 32-bit float back a few years ago. And then of course WAVpack was a no-brainer as pretty much the only way to losslessly compress a 32-bit float without esoteric hassles.

I would do 48 kHz 32-bit float audio files as my main format, since the higher sample rate of the DAW audio engine sometimes makes filters sound a bit better, but most of the time I can't hear the difference so I just stick with 44.1kHz. And the filesizes can get big!


By the way, OcenAudio is a pretty nice freeware audio editor. It can open and save 32-bit (IEEE) floats. It's a pretty good program for other formats, and it can write to FLAC or OGG or a few other formats.
Go to the top of the page
+Quote Post
saratoga
post Nov 9 2013, 20:47
Post #21





Group: Members
Posts: 4715
Joined: 2-September 02
Member No.: 3264



QUOTE (Nystagmus @ Nov 9 2013, 15:34) *
And like some other folks, I like knowing that my tunes are rendered to their fullest bit-resolution quality possible using 32-bit float.


Technically 32 bit int is higher. 32 bit float is used for production because it won't clip like int. But if you're just storing music it doesn't really make sense.
Go to the top of the page
+Quote Post
kode54
post Nov 10 2013, 00:36
Post #22





Group: Admin
Posts: 4499
Joined: 15-December 02
Member No.: 4082



It still makes some sense if you haven't peak scaled your floating point WAV before converting to integer format. Or ReplayGain scan and apply, keeping the peak level at or below 1.0, in case whatever volume level it arrives at leaves peaks exceeding that.
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: 17th April 2014 - 08:28