44.1khz and 48khz same file size when exporting WAV -> mp3

I’m using Audacity to convert a WAV file to mp3. I’ve noticed that the resulting mp3 file has the same file size no matter if the WAV had a sample rate of 44.1khz or 48khz.

I was under the impression that a higher sample rate creates a larger file size. Why is this not the case for the Audacity export?

a higher sample rate creates a larger file size.

It does in straight, simple, uncompressed files. MP3 is not a straight, simple, uncompressed format.

I’d be a lot more comfortable if you analyzed each of your two original files and published the results before you tried the conversion.


I think this topic explains it: https://forum.audacityteam.org/t/bits-per-sample-conversion/63451/1

What do you want me to analyze? It’s not clear to me. I know the sample rate of each file.

My question in this thread was answered but as far as I know, it’s not related to the sample rate. Or am I missing something?

You’re missing something :wink:

When you make an MP3, you specify the “kbps” (kilo-bits per second).
Example: 128 kbps = 16 kilobytes per second. So regardless of anything else, 1 minute of audio at 128 kbps will be 60 x 16 = 96 kilobytes. (actually a tiny bit more due to metadata and other bits, but the audio data will be 96 kB for 1 minute of audio).

This means that 48khz only results in a larger file size in WAVs (and probably some other file types, but not mp3s)?
What I am uploading is a podcast and I see a lot of discussion about 44.1khz vs 48khz regarding file size. Does this mean they are not talking about mp3s?

Are there any quality differences between 44.1 and 48khz in mp3s and if yes, how are they conveyed if the file size is the same? Somewhere this additional information has to be stored, right?

Would you recommend using 44.1 or 48khz for mp3s?

Yes, but a qualified “yes”.

MP3s may be “Constant BIT Rate” (CBR) or “Variable Bit Rate” (VBR)*

In the case of CBR MP3, the size of the audio data is totally predictable: “kbps x duration in seconds = size of audio data”

In the case of VBR MP3, the size of the audio data is less predictable as the encoder aims for a specified quality rather than an exact bit rate. The “Standard quality” setting usually gives around 170 to 210 kbps, but more complex sounds will tend to be bigger than simple sounds. The MP3 encoder automatically adjusts the bit rate to achieve the desired quality.

* There’s also “Average Bit Rate” MP3s, but they are fairly rare these days

I’ve not seen the discussion, so I don’t know what they are referring to.
Podcasts generally use CBR MP3 because some MP3 players have buggy support for VBR, which is a shame because VBR is better. Podcasters will want maximum compatibility.

For medium bit-rate CBR MP3, (example, 128 kbps), audio that is 44.1 kHz may give slightly better sound quality than audio that is 48 kHz.
Standard quality VBR MP3 will often produce smaller files with 44.1 than 48 kHz.

For podcasts, I’d recommend 44.1 and CBR.

Interesting, this is not very intuitive. But I think I noticed that when I compared the 44.1khz and 48khz mp3s. I was confused because 48khz actually seemed to sound worse, particularly the intro soundtrack. But the intro soundtrack source file itself also had 44.1khz, so it makes sense to keep it at 44.1khz.

Regarding CBR and VBR, I’ve also noticed this because VBR made scrubbing not work properly in online players and the total episode length was off. This is what led me down this rabbit hole :slight_smile:

Thank you very much for the clarification, that was very helpful!

The difference should be pretty small, but there’s no benefit in using 48 kHz if the final format will be 128 kbps MP3.

Yes, that’s the kind of bug that is still quite common, despite the fact that VBR has been around for over a decade.

Thank you for your kind help! This board is amazing!