Best Practices for Structure of a Project

Hi,

I been viewing some online videos about digital audio and what I’m learning is raising questions in my mind about the approach I should be using for my audio projects – which are basically editing one hour seminars.

The organization that I edit audio for needs/wants the final audio to be at 44.1 kHz and at 16 bit PCM but they usually send it to me at 32 kHz and 32-bit float.

Should I resample the audio and change to 16 bit PCM when I start the project or should I wait until after all the editing is completed and do the conversion at the time I export the completed files? I’m thinking that the sample rate should be changed right away since the higher sample rate would provide more data points for editing and effects. But since 32 bit is more accurate than 16 bit it seems I wouldn’t want to change from 32 to 16 until the end of the project.

I normally set the Project Rate at 44.1 kHz and 16 bit and so new tracks I generate during editing will be at those rates. If I copy and paste audio from a 32 kHz/32 bit track to a 44.1 kHz/16 bit track, I believe the audio format is automatically converted - correct? And since when I export the audio it will be exported at the Project Rate and I have the option at that time to choose 16 bit - I believe that all the format changes could be accomplished at export time - correct?

I’m wondering also if there any issues that could arise from having different audio formats within the same project? And what is the best point in time to handle the conversion from 32/32 to 44.1/16?

Just looking for best practices to structure these projects.

Thanks

Mike

With a seminar the weak link is most likely acoustic noise and overall recording quality. And in general the digital-side of things isn’t “the problem”.

By default, Audacity works “internally” at 32-bit floating point. If you are save your “project” it’s also saved in floating-point. There are advantages to “processing” in floating point. If you are exporting temporary intermediate WAV files you can use 24 or 16-bit (regular integer), or you can use 32-bit floating point. The main advantage to floating-point for temporary files is that it effectively has no upper limit so it can go over 0dB without clipping. (There is also no lower limit). Of course, your files will be proportionally larger if you use 32-bits.

As a practical matter, 16-bits is almost always better than human hearing so that’s good enough for just about anything. (Just for reference, pro studios generally use 24-bit/96kHz for recording and file storage with processing done in floating point.)

they usually send it to me at 32 kHz and 32-bit float.

That’s odd… Is that a WAV file?

I normally set the Project Rate at 44.1 kHz and 16 bit and so new tracks I generate during editing will be at those rates. If I copy and paste audio from a 32 kHz/32 bit track to a 44.1 kHz/16 bit track, I believe the audio format is automatically converted - correct?

It will be converted to your project rate when you export so that’s not necessary. Audacity might “work faster” if your files and project rate match, but I’m not sure.

And since when I export the audio it will be exported at the Project Rate and I have the option at that time to choose 16 bit - I believe that all the format changes could be accomplished at export time - correct?

Yes, you choose the bit depth when you export. The conversion from 16 or 24-bits to 32-bit floating point and back is lossless.

I’m wondering also if there any issues that could arise from having different audio formats within the same project? And what is the best point in time to handle the conversion from 32/32 to 44.1/16?

It’s no problem. Everything will be converted to your selected Project Rate when you export. It’s “good practice” to avoid unnecessary conversions so it’s best to convert once. But practically speaking, you’re probably not going to hear any difference with multiple conversions back-and-forth.

It will make very little difference whether you change the sample rate at the start or at the end. I’d change it at the start so that I don’t forget later.

The sample format is more important. Use “32-bit float”.

As DVDdoug mentioned, Audacity uses 32-bit float internally, which provides exceptionally precise processing, and also protects against the possibility of “clipping” while working on the audio. (“Clipping” occurs with integer formats, such as 16-bit, if the amplitude goes over 0 dB. 32-bit float format can handle signals over 0 dB, so you just have to ensure that the waveform is below 0 dB when you export).

Audacity uses 32-bit float by default, and this is virtually always the best choice.

Thanks DVDdoug and Steve,

Your feedback is very helpful.

The organization that I edit seminars for is very small and doesn’t have any full time person allocated to their “digital audio/video department”. So I think the instructions that have been put together to share with the volunteer audio editors have been based on feedback from various volunteers in the past (like me :slight_smile: ). So it tends to be a hodgepodge of very basic steps that allow the audio to be processed but that do not utilize Audacity’s features to any great extent.

For example, to change the sample rate, the instructions they provided said to “click the new rate in the Rate field in the Audio Track Drop Down Menu - but by changing the rate, the length of the recording will change, altering the sound of the voices. You will now need to adjust the speed of the recording in order to rectify this …”. So that is what I was doing to start with until I stumbled across the Track > Resample option. At any rate, I’m hoping to re-write the instructions sometime soon and, participating in this Forum has been very helpful in that regard.


They export the audio file from the video of the seminar and it comes in .M4A format. I believe that is a “lossy” format and so maybe the concern I raised with using 32-bit won’t really have any impact on these files. They do send .WAV periodically and I should probably mention that they should be sending .WAV on a regular basis.

Thanks again,

Mike