How do I know what Bit Depth my project / track is and if I should apply Dithering when exporting?

[Richard Burns]

Hi,

 

I'm completely new to this community and forum, so thanks in advanced for taking the time to read this.

I've hit a wall with my Cakewalk progress and I cannot seem to find a solid or straightforward answer to an issue I have, even after hours of looking.

Basically, I am at a stage where I am ready to export my project as a WAV file. Within the Export Audio screen, I have set the Sample Rate at 48k, as this is what the project was set to within preferences (48k was chosen due to track eventually being used with a video).

However, when it comes to choosing the Bit Depth within the Export Audio screen, I’m completely stuck. I initially thought there was only one place you set the bit depth for a project and it was within "driver settings" within preferences (much like the sampling rate I have just mentioned). However, I discovered yesterday that there are 3 different instances where bit depth can be set within a project which are as follows:

 

Audio Driver Bit depth within driver settings (this is fixed at 24 bit for me due to the ASIO I assume)

Record Bit Depth within preferences > Audio Data (this is set to 16 bit which is a shame because everything already recorded is now at 16 bit and not 24, but at least I will know for next time)

Render Bit Depth within preferences > Audio Data (this is set to 32 bit).

 

So my question is: which one of these do I go off to know if I need to apply dithering or not when exporting the track? How do I know if I am exporting the project / track at a lower bit rate to what it is set to?

So should I export at 16 bit (as that’s what the tracks are recorded at)? Or should I export at 24 bit (as that is what the audio driver is set to). Or, should I export to 24 bit and apply dithering because the Render Bit depth is set at 32? Are elements of my project going to be at 32 bit due to this, meaning if I export to 24 bit without applying dithering, I could get unwanted artefacts or a bad sound?

So I really don't know what to go off for this, or how to make a decision on what to do or select on the export process.

 

I hope this all makes sense and I would be very grateful of and explanation of these different bit depths and which one I should be looking at and / or go off to resolve the problem just described.

 

Thanks Again and Kind Regards.

 

[scook]

Record bit depth is used for ... recording new clips. Usually, this should be set the same as the audio driver bit depth.

Render bit depth is used when the DAW creates clips by some type of processing such as freezing, bouncing and exporting clips. IOW, everything except recording and importing.

Dithering is usually performed when going from a higher bit depth to a lower one such as 32bit to 16bit. So by this "rule" going from 32bit to 24bit one would apply dithering. That said, there is practically no audible difference creating 24bit files from 32bit (or 64bit) whether dithered or truncated. Do apply dithering when going from anything higher than 16bit down to 16bit. This often does make a noticeable difference.

 

[scook]

40 minutes ago, Richard Burns said:

So should I export at 16 bit (as that’s what the tracks are recorded at)? Or should I export at 24 bit (as that is what the audio driver is set to). Or, should I export to 24 bit and apply dithering because the Render Bit depth is set at 32? Are elements of my project going to be at 32 bit due to this, meaning if I export to 24 bit without applying dithering, I could get unwanted artefacts or a bad sound?

Unless you plan on taking the export into another tool such as Ozone or and audio editor for additional processing, export to the bit depth you plan on using while listening. 

If exporting for additional processing, export at the render bit depth.

Dithering, if done at all, should happen one time, when creating the final version of the file.

[Richard Burns]

3 hours ago, scook said:

Record bit depth is used for ... recording new clips. Usually, this should be set the same as the audio driver bit depth.

Render bit depth is used when the DAW creates clips by some type of processing such as freezing, bouncing and exporting clips. IOW, everything except recording and importing.

Dithering is usually performed when going from a higher bit depth to a lower one such as 32bit to 16bit. So by this "rule" going from 32bit to 24bit one would apply dithering. That said, there is practically no audible difference creating 24bit files from 32bit (or 64bit) whether dithered or truncated. Do apply dithering when going from anything higher than 16bit down to 16bit. This often does make a noticeable difference.

 

 

3 hours ago, scook said:

Unless you plan on taking the export into another tool such as Ozone or and audio editor for additional processing, export to the bit depth you plan on using while listening. 

If exporting for additional processing, export at the render bit depth.

Dithering, if done at all, should happen one time, when creating the final version of the file.

 

Thanks for taking the time to respond to this. Ok, so from what you have said, am i right in assuming the following:

 

Even if the record & audio driver  bit depth is set at 16 or 24 Bit,  the moment you apply any form of processing or editing in the daw (automation, adding an fx, etc) it's going to be treated as 32 bit, due to the Render Bit depth. 

By extension of this, am i right in assuming i should be looking at the Render Bit Depth to know what my overall mix / project is? This is important to know due to knowing if i need to apply Dithering or not to my final "master" export.

 

Also, should i leave my Render Bit Depth at 32 bit and not just set all of these 3 settings at 24 bit?  I've heard having the render bit depth at 32 is beneficial during the mixing process?

 

Lastly, I've found that there seems to be a massive debate on whether you need to dither from 32 bit to 24 and this was reflected in your post. So In your experience, would you suggest not dithering at all in this instance and only when going down to 16 Bit. I've heard a range of things on this subject, such as it causing more harm than good when dithering from 32 bit to 24 but i could use some guidance on it.

[scook]

42 minutes ago, Richard Burns said:

Even if the record & audio driver  bit depth is set at 16 or 24 Bit,  the moment you apply any form of processing or editing in the daw (automation, adding an fx, etc) it's going to be treated as 32 bit, due to the Render Bit depth. 

The audio engine runs at 32 or 64 bit depending on the 64bit Double Precision Engine setting in preferences. When the DAW creates files other than recording or importing, it uses the Render Bit Depth.

42 minutes ago, Richard Burns said:

By extension of this, am i right in assuming i should be looking at the Render Bit Depth to know what my overall mix / project is? This is important to know due to knowing if i need to apply Dithering or not to my final "master" export.

Unlike sample rate, projects can have clips at a variety of bit depths. As mentioned before, the Render Bit Depth is used any time files are written to disk except when recording or importing. Recording and importing have their own settings on the same preference page.

42 minutes ago, Richard Burns said:

Also, should i leave my Render Bit Depth at 32 bit and not just set all of these 3 settings at 24 bit?  I've heard having the render bit depth at 32 is beneficial during the mixing process?

Use either 32 or 64 for Render Bit Depth. The default is 32.

The Record Bit Depth should be set to agree with the Audio Driver Bit Depth. By default, Record Bit Depth is 16 but most ASIO drivers use 24. Some interfaces with built-in DSP are locked at 32.

I leave the Import Bit Depth at "Original." Cannot think of a reason to change this. If necessary, the sample rate will be altered to the project sample rate but the DAW can handle clips at different bit depths.

42 minutes ago, Richard Burns said:

Lastly, I've found that there seems to be a massive debate on whether you need to dither from 32 bit to 24 and this was reflected in your post. So In your experience, would you suggest not dithering at all in this instance and only when going down to 16 Bit. I've heard a range of things on this subject, such as it causing more harm than good when dithering from 32 bit to 24 but i could use some guidance on it.

Dithering 32/64 floating point to 24bit may matter for some genres and listening environments but I have my doubts. That said, if this is the final export, dithering won't hurt. Try it and see. In some cases it may not even matter when going to 16bit. It really depends on the material and the final audio format. I rarely export 16bit anymore. When I do, I use dithering. 

Bottom line, dithering should be done at most one time. It is most often used when creating 16bit files but there is no harm using anytime the bit depth is being reduced.

[Richard Burns]

39 minutes ago, scook said:

The audio engine runs at 32 or 64 bit depending on the 64bit Double Precision Engine setting in preferences. When the DAW creates files other than recording or importing, it uses the Render Bit Depth.

 

 

Would you possibly be able to give me an example of files that the DAW can create in a project that would use the Render Bit Depth? I understand what you're saying but  I think an example would help me to get my head around how it works.

 

43 minutes ago, scook said:

Unlike sample rate, projects can have clips at a variety of bit depths. As mentioned before, the Render Bit Depth is used any time files are written to disk except when recording or importing. Recording and importing have their own settings on the same preference page.

 

  

I see... So as there are multiple bit depths used within the project, should i be going off the highest one that will be used in the project to know if i will be exporting  the final version at a lower / reduced Bit rate overall? So for example, if i want to export the final version at 24 Bit depth, i know that the project will contain some elements with bit depths at 32, as that is what the Render Bit Depth is set at. So by this logic, i would be exporting the final version at a lower bit depth. Is this assumption correct?

[scook]

6 minutes ago, Richard Burns said:

Would you possibly be able to give me an example of files that the DAW can create in a project that would use the Render Bit Depth? I understand what you're saying but  I think an example would help me to get my head around how it works.

• Freezing and Bouncing tracks - both of these apply effects and generate audio from MIDI using softsynths. Bounce can also mix down multiple tracks to a single track.
• Destructive edits - any time an audio clip is edited in a way causing it to be written to disk such as shortening a clip and throwing away the unused part (aka applying trimming). Destructively applying plug-ins. Normalizing a clip.
• Exporting files

14 minutes ago, Richard Burns said:

I see... So as there are multiple bit depths used within the project, should i be going off the highest one that will be used in the project to know if i will be exporting  the final version at a lower / reduced Bit rate overall? So for example, if i want to export the final version at 24 Bit depth, i know that the project will contain some elements with bit depths at 32, as that is what the Render Bit Depth is set at. So by this logic, i would be exporting the final version at a lower bit depth. Is this assumption correct?

Internally, the DAW reads all the audio clips and gets all the audio data from the synths and mixes this together using 32 or 64bit math. This data is sent to the export process. By default, export uses the Render Bit Depth to write out the file; however, this may be overridden in the export dialog. If exporting a 24bit file, the extra info in the data stream has to be dealt with in some fashion. This means either truncating or dithering.

[Richard Burns]

7 minutes ago, scook said:

• Freezing and Bouncing tracks - both of these apply effects and generate audio from MIDI using softsynths. Bounce can also mix down multiple tracks to a single track.
• Destructive edits - any time an audio clip is edited in a way causing it to be written to disk such as shortening a clip and throwing away the unused part (aka applying trimming). Destructively applying plug-ins. Normalizing a clip.
• Exporting files

 

This makes sense to me now. Thanks for that

8 minutes ago, scook said:

Internally, the DAW reads all the audio clips and gets all the audio data from the synths and mixes this together using 32 or 64bit math. This data is sent to the export process. By default, export uses the Render Bit Depth to write out the file; however, this may be overridden in the export dialog. If exporting a 24bit file, the extra info in the data stream has to be dealt with in some fashion. This means either truncating or dithering.

 

Right ok. So basically the overall project will be mixed together at either 32 bit or 64 bit, depending on the 64bit Double Precision Engine setting in preferences. So this means on the final export / file it will always be lower if choosing to  export to 24 bit, meaning there is no harm applying Dithering. Would this be correct?

 

If so then it sounds like the final export will always be lower if exporting to 24 bit.

 

 

[scook]

Exporting at 24bit is less than the internal mix. Try with dithering and without. Try the different dithering algorithms. See if it makes a difference with your own projects. The "rule" is always dither when lowering bit depth but, it often makes no audible difference.

[Richard Burns]

43 minutes ago, scook said:

Exporting at 24bit is less than the internal mix. Try with dithering and without. Try the different dithering algorithms. See if it makes a difference with your own projects. The "rule" is always dither when lowering bit depth but, it often makes no audible difference.

I think i understand this now so thanks a lot for this. It sounds like if i export to 24 bit with or without dithering, it won't make a massive difference but i will experiment as you have suggested.

 

Doing the final export at 24 bit won't impact the audio that was recorded at 16 bit will it? As mentioned this wasn't set to match the audio driver bit depth (which is 24 bit) as i didn't know it was an option. From my understanding though, it won't do any harm if the export is higher than this  particular setting and i will change the settings of this for future projects now that i know.

[scook]

Converting 16bit files to 24bit files does not do a thing to the sound of the files. 

[Richard Burns]

Thanks again for all your help and advise, it really is appreciated.?

[Bruno de Souza Lino]

All claimed differences in regards to dithering are not audible by human ears. The spectrum is different and differences can be measured. In practice, truncating or dithering makes no difference and nobody will know if you did X or Y. Unless someone invents a box that you can adjust from zero to a lot of dither, they're just parroting stuff they heard somewhere else.