Audacity working internally in 32bit format

Feedback and Reviews for Audacity 2.x

This board is ONLY for general feedback and discussion about Audacity 2.X.

If you require help, or think you have found a "bug", please post on the forum board relevant to your operating system.
Windows
Mac OS X
GNU/Linux and Unix-like

Audacity working internally in 32bit format

Permanent link to this post Posted by mfarooq7897 » Sun Apr 24, 2016 6:58 am

Ladies and Gentlemen,

I understood from my previous post the cause of some noise being added in waveforms every time they are exported in a non 32bit format. The previous post cold be seen on the link at the bottom. You don't have to read it and heres the summary of that post:
Some noise is added to the waveforms while exporting because of dithering. Since Audacity works internally in 32bit, everytime a lower bitrate is exported, it incurs harmonic distortion and to cope with it, dithering is done which ends up adding noise.

My question is:
What if audacity is made to read, and work internally, in the file's actual format. This would not require need for dithering.
My guess is that, everytime a file of bitdepth lower than 32 is read, its bit depth is raised to work internally. This raising of bitdepth (for audacity to work with it) and reduction (because I want to save it as, say, 16bit) necessitates dithering which in turn compromises the integrity of waveform.

Sorry for being dim here.

Thanks


viewtopic.php?f=50&t=91028&p=303410#p303410
mfarooq7897
 
Posts: 6
Joined: Mon Oct 26, 2015 8:16 pm
Operating System: Please select

Re: Audacity working internally in 32bit format

Permanent link to this post Posted by Gale Andrews » Sun Apr 24, 2016 12:06 pm

mfarooq7897 wrote:I understood from my previous post the cause of some noise being added in waveforms every time they are exported in a non 32bit format. The previous post cold be seen on the link at the bottom. You don't have to read it and heres the summary of that post:
Some noise is added to the waveforms while exporting because of dithering. Since Audacity works internally in 32bit, everytime a lower bitrate is exported, it incurs harmonic distortion and to cope with it, dithering is done which ends up adding noise.

Dither is added when you downconvert the bit depth (sample size). Not when you downsample by reducing the sample rate.

If you are talking about WAV, it does technically have a bit rate which is number of channels * bit depth * sample rate. So if dither is applied it is because you are reducing the WAV bit rate by downconverting, but you could reduce the bit rate by downsampling and there would be no need for dither.

mfarooq7897 wrote:My question is:
What if audacity is made to read, and work internally, in the file's actual format. This would not require need for dithering.
My guess is that, everytime a file of bitdepth lower than 32 is read, its bit depth is raised to work internally. This raising of bitdepth (for audacity to work with it) and reduction (because I want to save it as, say, 16bit) necessitates dithering which in turn compromises the integrity of waveform.

Raising the bit depth does not require dither, only reducing the bit depth.

You can set the bit depth each time in Quality Preferences according to the bit depth of the file, but that only holds if you don't change the sample amplitudes by running an effect. Cutting some audio out won't change the sample amplitudes, so if you import 16-bit audio at 16-bit resolution, and only cut, the audio remains at 16-bit resolution all the time.

If you apply an effect like Fade or Amplify, then the 16-bit audio is upconverted to 32-bit float for processing then downconverted to 16-bit and dithered when returned to the track.

Working internally in 16-bit would get very noisy if you amplified the audio many times. You can search on the Forum for some experiments that prove that.


Gale
________________________________________FOR INSTANT HELP: (Click on Link below)
* * * * * Tips * * * * * Tutorials * * * * * Quick Start Guide * * * * * Audacity Manual
Gale Andrews
Quality Assurance
 
Posts: 26087
Joined: Fri Jul 27, 2007 12:02 am
Operating System: Windows 10

Re: Audacity working internally in 32bit format

Permanent link to this post Posted by kozikowski » Sun Apr 24, 2016 1:02 pm

32-floating doesn't overload. If you apply an effect that causes the show to be louder than 0dB, a 16-bit show would become permanently damaged. A sequence of effects in a chain would damage the show if any of the steps in the middle caused the sound to exceed 0dB.

A 32-floating show is simply reduced in volume back to normal using any volume tool. In the case of a chain, the final show is determined by all of the tools, not restricted by any one step in the middle.

Koz
kozikowski
Forum Staff
 
Posts: 40209
Joined: Thu Aug 02, 2007 5:57 pm
Location: Los Angeles
Operating System: OS X 10.9 Mavericks

Re: Audacity working internally in 32bit format

Permanent link to this post Posted by steve » Sun Apr 24, 2016 2:34 pm

mfarooq7897 wrote:What if audacity is made to read, and work internally, in the file's actual format. This would not require need for dithering.

If, for example, Audacity was made to work internally in 16-bit from 16-bit audio and exporting as 16-bit, then you are correct that dither would not be required and not used. However, the audio quality would suffer when any kind of processing is performed. For example, if you amplify the audio, then in virtually all cases (except for exact doubling, quadrupling etc) the sample values would need to be rounded to the nearest 16-bit integer value. This rounding would manifest as noise, and would be audibly worse than processing as 32-bit float + dither.

The main benefits of processing as 32-bit float (compared to processing in a lower "integer" sample format) are:
* The data does not clip at 0 dB
* Less noise
* Less distortion
* Greater dynamic range
9/10 questions are answered in the FREQUENTLY ASKED QUESTIONS (FAQ)
steve
Site Admin
 
Posts: 47011
Joined: Sat Dec 01, 2007 11:43 am
Operating System: Linux *buntu

Re: Audacity working internally in 32bit format

Permanent link to this post Posted by Gale Andrews » Sun Apr 24, 2016 2:45 pm

steve wrote:The main benefits of processing as 32-bit float (compared to processing in a lower "integer" sample format) are:
* The data does not clip at 0 dB
* Less noise
* Less distortion
* Greater dynamic range

... and another is that most modern computers are faster performing calculations in floats than integers.


Gale
________________________________________FOR INSTANT HELP: (Click on Link below)
* * * * * Tips * * * * * Tutorials * * * * * Quick Start Guide * * * * * Audacity Manual
Gale Andrews
Quality Assurance
 
Posts: 26087
Joined: Fri Jul 27, 2007 12:02 am
Operating System: Windows 10


Return to Audacity 2.x Feedback and Reviews



Who is online

Users browsing this forum: No registered users and 2 guests