"Bits per sample" conversion

Hey,
I export a WAV file from Premiere Pro (because it also has a video version). When I check this file in VLC player, it says that it has a “bits per sample” value of 16.
I then import this WAV file into Audacity to export it as an mp3. But the “bits per sample” value of that mp3 is now 32. Am I wasting disk space with that? Should I let Audacity export with the same 16 value and if yes, how?

By default, Audacity automatically converts audio on import to “32-bit float” format. This has several benefits:

  1. Audacity uses 32-bit float internally, so working in 32-bit float avoids unnecessary conversions.

  2. 32-bit float is able to go over 0 dB, so it eliminates the risk of clipping while working on a project (though you do have to ensure that the audio does not exceed 0 dB before you export).

  3. Modern computers are generally a bit more efficient when working with 32-bit float rather than 16-bit integer.

  4. During processing, 32-bit float is much more precise than 16-bit.

In short: Work in 32-bit float, Export in whatever format you need.

Thank you for your answer!
Are files exported with 32-bit bigger than 16-bit? And how do I change to 16-bit? I tried via the preferences but it had no effect on the exported mp3.

MP3 does not use “bits per sample”, it uses “bits per second” (more usually described as kilo-bits per second “kbps”).

Yes. For uncompressed formats such as WAV, a 32-bit file is double the size of a 16 bit file, but I am not suggesting that you export as 32-bit.


When you export as WAV, the default is 16-bit. See here where it says:
“Encoding: Signed 16-bit PCM”


But when I open the media info of the mp3 file in VLC player, it says “Bits per sample: 32”:
https://imgur.com/f3fDV4o

What’s the meaning of that? The same value in the WAV file (before importing it into Audacity and then exporting an mp3) is 16.
I am afraid of accidentally bloating up the file size artificially

I don’t know but VLC may be decompressing it to 32-bits just like Audacity does. The MP3 format (and the other popular lossy formats) doesn’t contain individual sample-values. The samples are “re-created” when it’s decoded/decmpressed and played, but since it’s lossy compression the samples are not identical to the uncompressed original samples.

The bitrate in kilo_bit_s-per-second is the “file size” (not including embedded album artwork other metadata). There are 8 bits in a byte so you can divide by 8 to get the file size in kilobytes per second.

So with the default settings, Audacity is not adding any unnecessary file size bloat when converting an wav to mp3? I can ignore that value VLC player shows me? I’m clear on the kbps part (I’ve picked 128kbps on purpose).

When you export as 128 kbps MP3, the file size should be the same regardless of how many bits per sample in Audacity.
“128 kbps” means that each second of audio uses 128 kilo-bits, so the file size will be 16 kB per second of audio.

Awesome, thank you very much for the clarification!

Think about that… 128kbps is 16K (bytes) per second so a 1 minute file should be about 960K.