Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Which dithering to use in Audacity for 16 bit audio. (Read 14327 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Which dithering to use in Audacity for 16 bit audio.

Hi everyone. I've searched for topics on the same subject, but didn't find a complete answer to my question.

I'm using Audacity for editing audio, but only CD quality audio (16/44.1). I just found out that the default setting is to import the audio at 32 bit, and converting it back to 16 bit when exporting. I found out that there was a little bit of noise applied throughout the track when opening it again, even to complete silence parts of the track (that I added in Audacity). When searching the internet for solutions, I found out this was probably caused by dithering.
I usually import and export the tracks multiple times, so I'm afraid the dithering will 'build up' and affect the sound quality of the audio.

I read that dithering could be set to 'none' when one only uses 'basic editing' such as cutting and pasting and when using 16 bit audio. Does this apply to my case? I don't use any effects, sometimes however I might amplify a track and I often use fade-ins and -outs. Would this be considered 'basic editing'?

And what would be the 'cons' of not using dithering? I read about that I might hear distortions of that that certain frequencies might be added or removed...or does this only apply when downsampling 'real' 32 bit audio to 16 bit, instead of '16 bit imported as 32 bit'?

I hope somebody here can help me out. I realize I asked quite a lot of questions.

Thanks in advance!

Which dithering to use in Audacity for 16 bit audio.

Reply #1 –
Quote from: Luap on
I usually import and export the tracks multiple times, so I'm afraid the dithering will 'build up' and affect the sound quality of the audio.

You might consider saving it as an Audacity project, that way it is kept at 32-bit. If that's not an option, consider saving to 24-bit until after the last edit.
Music: sounds arranged such that they construct feelings.

Which dithering to use in Audacity for 16 bit audio.

Reply #2 –
Hello Luap,

I recommend this video. It might clear up a few things about digital-audio including dither.

Cheers
Chrisitian

Which dithering to use in Audacity for 16 bit audio.

Reply #3 –
Thanks for the replies so far. A very interesting video!

However, I don't have all the technical knowledge required to judge what the best thing is to do in my situation.

I tried to import the tracks as 16 bit to begin with; that didn't make a difference with dither still 'on', noise was still being applied when exporting.

Would it be best to just keep editing in 32 bit, but set dither to 'none' in my case? Or does fadeing in and out need some 'correction' from the dithering?

Thanks!

Which dithering to use in Audacity for 16 bit audio.

Reply #4 –
I remember having had a similar problem with Audacity. IIRC importing a wav file originating from CD and immediately exporting it without any audio editing changed the file.
I solved the issue by changing the Preferences, High Quality Sinc interpolation and No Dither, IIRC.
Sorry for not being sure about the details.
lame3995o -Q1.7 --lowpass 17

Which dithering to use in Audacity for 16 bit audio.

Reply #5 –
Quote
Would it be best to just keep editing in 32 bit, but set dither to 'none' in my case? Or does fadeing in and out need some 'correction' from the dithering?
"None" should be fine. 

The theory is you should dither once when you downsample and dither as the last step (or nearly the last step) after all processing and editing has been done. 

Fading-in and fading-out might benefit from dither since you are moving some information "down below" the original 16-bits.    The classic example is when adding reverb, where you want the reverb-tail to smoothly "dither away", rather than fading to a point where it falls below 1-bit and suddenly goes dead-silent.   

The real answer is, do whatever sounds best to you!

But under normal listening conditions, you shouldn't be able to hear dither at 16-bits.  That is, when you crank-up volume loud enough to hear dither (or the effects/benefits of dither) during fade-in or fade-out, the overall main program volume of will be so high that the sound will be badly distorted and/or ear-damaging.    (You can probably hear it if you downsample to 8-bits, but I don't know if Audacity will dither an 8-bit sample.)

P.S.
Nobody ever listens to a recording and says, "You forgot to dither", or "You used the wrong dither".  So, try not to loose too much sleep over it...  Focus on the stuff that really makes a difference.

Which dithering to use in Audacity for 16 bit audio.

Reply #6 –
Quote from: halb27 on
I remember having had a similar problem with Audacity. IIRC importing a wav file originating from CD and immediately exporting it without any audio editing changed the file.
I solved the issue by changing the Preferences, High Quality Sinc interpolation and No Dither, IIRC.
Sorry for not being sure about the details.


That seems to be very logical. However, in my situation I do alter the audio file, by editing and fading. This is from the Audacity Wiki (EDIT: from the 'Dither' page):

"Here's the exception to the rule: If you have recorded in 16-bit and are only doing simple editing (cut, delete, paste, trim...) and not doing any processing (amplify, equalize, frequency filter....) then for highest accuracy dither can be set to "none". In this case, because there are no 32-bit operations prior to export there is no benefit to using dither. Exporting a 16-bit track to 16-bit with dither set to "none" will be lossless. The same applies if exporting from a 24-bit track to an uncompressed 24-bit file format with dither disabled. "

Now the question is, is fading (in or out) some sort of amplifying, and thereby processing (which might need some dithering, according to the text above)?
Or would it just be considered basic editing.

And also, does it make sense to import and edit an original 16 bit track in 32 bit? Does that give a better quality? Or doesn't it make a difference when I would just import and edit the 16 bit track in 16 bit in Audacity?

Which dithering to use in Audacity for 16 bit audio.

Reply #7 –
Quote
And also, does it make sense to import and edit an original 16 bit track in 32 bit? Does that give a better quality? Or doesn't it make a difference when I would just import and edit the 16 bit track in 16 bit in Audacity?
It shouldn't matter.  I believe Audacity (like most audio editors) works in 32-bit floating-point internally. 

Floating point makes the DSP easier, it  makes it impossible to clip Audacity internally, and you don't loose information/resolution (internally) if you reduce the volume.  And if you don't change anything, you can go from 16-bit (or 24-bit) integer to 32-bit floating-point and back losslessly.

Which dithering to use in Audacity for 16 bit audio.

Reply #8 –
Quote from: Luap on
...
Now the question is, is fading (in or out) some sort of amplifying, and thereby processing (which might need some dithering, according to the text above)?
Or would it just be considered basic editing.

And also, does it make sense to import and edit an original 16 bit track in 32 bit? Does that give a better quality? Or doesn't it make a difference when I would just import and edit the 16 bit track in 16 bit in Audacity?


Fading is processing (amplifying). Best practice is to dither afterwards. Try hard to avoid doing any processing in 16 bit mode.

The advice you've been given so far is all good.

- If you are only cutting and pasting (removing samples, moving samples around, adding digital silence), you are not actually changing the value of any of the samples. You don't need to dither when saving. Although in theory you could do such editing in 16 bit mode, you may as well use the default 32 bit mode in Audacity - it won't hurt, and it may help if you decide you do want to do addiitonal processing.

- If you do any processing that changes the value of any of the samples, such as fade or normalise or equalise etc, you should in most cases dither when saving. Never do such processing in 16 bit mode - always use 24 bit or 32 bit.

- If you plan to make additional changes to a track at a later time, save a copy in Audacity project format. If you must save only a 16 bit version, use TPDF (Triangle) dither rather than Shaped. This causes less damage when multiple layers of dither are applied.

This can be summed up by a few simple rules:
- Always process in 24 bit or 32 bit mode.
- Always dither when reducing from a higher bit count to a lower bit count. (32 -> 24, 32 -> 16 etc).
- Try to dither only once, when writing the final file. This implies keeping intermediate saves in 24 or 32 bit format.

Experts sometimes break one or more of these rules, but they know when they can get away with it. Following the rules will keep you out of trouble until you become an expert.

Regards,
   Don Hills
"People hear what they see." - Doris Day

Which dithering to use in Audacity for 16 bit audio.

Reply #9 –
Quote from: splice on
- If you do any processing that changes the value of any of the samples, such as fade or normalise or equalise etc, you should in most cases dither when saving. Never do such processing in 16 bit mode - always use 24 bit or 32 bit.

- If you plan to make additional changes to a track at a later time, save a copy in Audacity project format. If you must save only a 16 bit version, use TPDF (Triangle) dither rather than Shaped. This causes less damage when multiple layers of dither are applied.


Thanks a lot for your reply, and all the help so far.

Does this mean I always have to use Triangle shaped dither, also when exporting the final 16 bit file?

EDIT: And also, might it be better to first take only the part I will fade out, and fade it out with dither, and then open it again and paste it into the full file which I will save without dither? That way only an very small part of my track will have the dither going on, only the faded part, the remainder of the musc will not have any dither.

Which dithering to use in Audacity for 16 bit audio.

Reply #10 –
I think you're getting a bit obsessed by this. The chances of audible differences at normal listening volumes are rather small.

That said, Cool Edit Pro / Audition will happily only dither the parts of the file you have changed, and leave the rest completely unchanged. So if you apply a fade in cool edit, the fade is dithered, and the rest of the file is not. I think that's how it works by default (though I've had this software so long that I have no real idea what the default settings were!). When you apply some change to a 16-bit file that would ideally require dither (i.e. gain change, not cut/paste), the calculation is carried out at 32-bit internally and then dithered down. Alternatively you can choose no dither at all, or converting the whole file to 32-bits while working, or ... as you wish.


If you are worried that someone might turn the volume up during the fade enough to be able to hear the least significant bit then there is no perfect solution when the output is 16-bits. Dithering everything adds noise throughout. Dithering only the fade adds noise which quite obviously disappears when you cut to the next track. Not dithering leaves harmonic distortion and grittiness at the end of the fade. However, normal people do not listen like this.

Cheers,
David.

Which dithering to use in Audacity for 16 bit audio.

Reply #11 –
Quote from: 2Bdecided on
I think you're getting a bit obsessed by this. The chances of audible differences at normal listening volumes are rather small.

That said, Cool Edit Pro / Audition will happily only dither the parts of the file you have changed, and leave the rest completely unchanged. So if you apply a fade in cool edit, the fade is dithered, and the rest of the file is not. I think that's how it works by default (though I've had this software so long that I have no real idea what the default settings were!). When you apply some change to a 16-bit file that would ideally require dither (i.e. gain change, not cut/paste), the calculation is carried out at 32-bit internally and then dithered down. Alternatively you can choose no dither at all, or converting the whole file to 32-bits while working, or ... as you wish.


If you are worried that someone might turn the volume up during the fade enough to be able to hear the least significant bit then there is no perfect solution when the output is 16-bits. Dithering everything adds noise throughout. Dithering only the fade adds noise which quite obviously disappears when you cut to the next track. Not dithering leaves harmonic distortion and grittiness at the end of the fade. However, normal people do not listen like this.

Cheers,
David.


You're absolutely right. However, I'm also remastering and fixing things in Audacity to create the best possible versions of some tracks, and I feel that my goal is a little nullified by all the dithering applied to most of tracks, and in many instances mutliple times.

But of course, there are very little situations in which one can hear the dither.

Which dithering to use in Audacity for 16 bit audio.

Reply #12 –
Depending on the music, I would get concerned by adding flat dither multiple times, or rounding many times. I'd be concerned at using extremely noise-shaped dither more than once.

As others have said, if you're doing lots of processing, just work at 32-bits until you're finished.

Many CDs are produced with less care than you're showing. I have CDs where several generations of dither are detectable, and where truncation is detectable. Not audible (and it wouldn't be the least of some of these CDs problems it is was), but detectable.

Cheers,
David.

Which dithering to use in Audacity for 16 bit audio.

Reply #13 –
Thanks to everyone for their answers and help, I've learned a lot.

I've decided to do a lot of the projects from scratch (it's not that much work) and keep them as an Audacity Project file in 32 bit, and save a 16 bit mixdown with Triangular dither of the latest version.
For projects with cutting and pasting only I'll probably make a mixdown without any dither.

Anyway, with all the 32 bit project files I can always change my mind later about any editing and mixdown decisions, without adding more layers of dither.

Which dithering to use in Audacity for 16 bit audio.

Reply #14 –
Quote from: 2Bdecided on
Many CDs are produced with less care than you're showing. I have CDs where several generations of dither are detectable, and where truncation is detectable. Not audible (and it wouldn't be the least of some of these CDs problems it is was), but detectable.


Hi 2Bdecided,

I'm just curious, could you list discs of yours that have inferrably been mastered with several generations of dither? (How careless and unprofessional, even if likely inaudible.) As for truncation, I suppose it'd have been unavoidable back in the 1980s.

Which dithering to use in Audacity for 16 bit audio.

Reply #15 –
For multiple dither, I was thinking specifically of this one...
http://www.hydrogenaudio.org/forums/index....st&p=143440
...IIRC it also seemed quite common on dance compilations in the early/mid 1990s, but I can't think of a specific example.

For truncation (in an era where people should have known better) I was thinking about a compilation disc from Deutsche Grammophon of their first "4D" tracks, where most tracks were faded out without dither. It went very strange and clicky at the tail of the fade-out. You'd never hear it at "normal" listening levels, but some of the tracks were already rather quiet (even by the standards back then) so you didn't need to turn the volume up much from a normal listening level to find it. I found it on my normal stereo before the days of ripping CDs, because I listened "too loud" to enjoy the supposedly superior hiss-free recording and long deep reverb tails of "4D" recording. The recordings were mostly very clean indeed - until the fade out.


As you say, it's unprofessional, but inaudible, so...

I'm sure we've all seen worse that's both unprofessional and audio - DRC, obviously, but also CDs mastered from audibly lossy sources.

Cheers,
David.

Which dithering to use in Audacity for 16 bit audio.

Reply #16 –
Interesting!. I have a Japanese Yes HDCD remaster or two, released circa 1998, that have 'strange clicky' sounds at very end, low level, during fadeout, I always wondered if that meant they had been sourced from vinyl ...maybe it's a truncation issue instead?

Which dithering to use in Audacity for 16 bit audio.

Reply #17 –
HDCD used the least significant bit (or was is 2 bits?) to store extra information in a 16-bit file. It sounds plausible that, if you amplified a fadeout, you might hear some of the HDCD-encoded information as noise. I'm not familiar enough with it to know if that would manifest as clicks, or something else.