Large files

[RodB]

Neophyte question: Saving/exporting MP3 produces a file bigger, sometimes much bigger, than the original file whether anything is done to the file or not. I open a 23MB mp3 file, delete half of it (I think...) and then when saved it's 25-27MB. Or I open a 5MB file, do nothing,and it gets saved/exported as a 25mb file. Is this normal? Anyway I can avoid it?

I'm trying to reduce a 25MB streamed audio (talking) mp3 file to 4 or 5MB to email a short (and smaller) clip.

[kozikowski]

Audacity doesn't edit MP3. It pulls the show into itself in a very high quality internal format, edits it, and then makes a whole new MP3 at the end. The output MP3 can be any size you want with changes to the output setup panel, but please note that you can never get back to your original size without additional bubbly, honky compression sound damage.

If all you're doing is hacking up an MP3 into segments, you might do well to use one of the editors that actually does edit MP3 natively like MP3Splt.

http://mp3splt.sourceforge.net/mp3splt_page/home.php

Koz

[RodB]

Thanks loads for the helpful info.

[Trebor]

I'm trying to reduce a 25MB streamed audio (talking) mp3 file to 4 or 5MB to email a short (and smaller) clip.

If it's only talking (not music) you could try changing the sample rate (project rate bottom left) from 44100Hz (default) to 22025Hz and exporting with a MP3 bit rate of 64Kbps. Also reduce the track to mono (if stereo) will reduce the file size.

Or you could share the unmodified 25Mb file via a file-sharing website like dropbox, sendspace, mediafire, etc who offer free accounts. You just email a link to the file for the OP to download.

[kozikowski]

Mono yes. That helps a lot.

However:

from 44100Hz (default) to 22025Hz

My money is that doesn't work. The quality of a compressed file depends on the quality of the input file and completely independent of the output filesize. The more you reduce the quality of the presented file, the worse the product is. The output filesize is purely dependent on the export settings.

You can't pre-distort an MP3 source file to make it smaller -- or any other compressed file type including video.

I need to submit this to Myth Busters.

Koz

[Trebor]

Mono yes. That helps a lot.

However:

from 44100Hz (default) to 22025Hz

My money is that doesn't work

I just tested that and you're right, the bit rate (kbps) defines the size of the mp3 file,
(I'm getting WAV and MP3 properties confused )
However there are less digital artifacts if you use a lower sample rate ...

''coming soon to a theatre near you'', original , [email protected] mp3, [email protected] mp3.flac

(305.26 KiB) Downloaded 118 times

[steve]

However there are less digital artifacts if you use a lower sample rate ...

That's because lower sample rate has a lower frequency bandwidth.
At 44.1 kHz (44100 Hz) sample rate, the highest possible audio frequency is 22050 Hz.
At 22050 Hz sample frequency the highest possible audio frequency is 11025 Hz.
The highest possible audio frequency at a given sample rate is half the sample rate - in practice it is a little lower.

MP3 encoding will attempt to reproduce as much of the audio spectrum as possible, with priority for the more easily audible frequencies. MP3 encoding always discards some of the audio data in order to achieve smaller file size/data bandwidth. The greater the compression, the smaller the file size, the less "bits per second" and the more data is discarded. For very high compression amounts, MP3 encoding will automatically resample to a lower rate and the user will see a pop-up dialogue screen to confirm the resampling.

For specific types of material it can be useful to pre-filter the audio so as to remove frequencies that you are not interested in. For example, for speech you can remove frequencies above about 7 kHz with very little loss to intelligibility, and then the MP3 encoder will not be trying to reproduce sounds higher than this.

Attached is "coming soon to a theatre near you" as mono 32kbps CBR MP3. The first half is directly from the sample from Trebor. The second has been pre-filtered at 6 kHz (6000 Hz) and has noticeably less "artefacts" without sounding too muffled.

[Trebor]

Attached is "coming soon to a theatre near you" as mono 32kbps CBR MP3. The first half is directly from the sample from Trebor. The second has been pre-filtered at 6 kHz (6000 Hz) and has noticeably less "artefacts" without sounding too muffled.

Is that "pre-filtered" via equalization (without changing the sample rate) , or via reducing the sample rate to 12000Hz ?

[steve]

Is that "pre-filtered" via equalization (without changing the sample rate)

Yes.