Changing tempo makes file three times bigger

Using Audacity 2.0.5 on a Windows 7 Thinkpad.

Backstory: I listen to a lot of podcasts while I drive, and I use Audacity to change the increase the tempo by about 35%. When I do that, I sometimes get a file that is three times as big as the original file: a 13Mb file becomes a 40Mb file. Can anybody explain why, or perhaps make a suggestion as to how to avoid that?

My goal is to be able to listen to as many podcasts in as short a time as possible, and they have to fit on my rather small, elderly .mp3 player.

Note: I tried reducing the sampling rate but about 10% of the time that results in extremely large files that are so slow I often can’t even make out what they’re saying - like a 78 playing at 33 rpm. (That last bit should tell you I’m a bit of a dinosaur, and I know next to nothing about how to do anything much to an audio file.)

I appreciate any input at all. Thanks.


Audacity will not edit an MP3 file. That may seem a little odd, but what it does is convert the show to a very high quality internal sound format in order to be able to perform effects and filters with little or no damage. This leaves it with the necessity of making a new MP3 when you’re done.

We warn people that the MP3 compression damage: bubbling, and honking and swimming doubles when you do this, but given what you’re doing, I don’t think it make any difference at all. There’s no good way to tell what the compression quality was on the original file, so if you always use the same Export compression, you could get wildly different file sizes.

You can develop a rhythm to your processing and know, through experimentation, that if you have a half-hour stereo show, you can use 128 MP3 quality and be presented with a certain size file. Note that if you don’t care about stereo, you can make a certain quality mono MP3 half the file size or double the quality. The two see-saw opposite each other.

You would think that if you restrict the fidelity or make the file muffled or AM Radio, you should be able to do much better, but that’s not the case. All that gives you is the same size MP3 with AM Radio on it. Compressors like MP3, AAC, M4A, etc. all work with the assumption that they’re starting with a perfect, high quality show and their actions and quality depend on that. If you don’t, then the results could be unpredictable.

Last note: The absolute minimum MP3 quality for Mono is 32 and Stereo is 64. Below that, a show will start to sound seriously compromised – and the effect changes slightly with the listener. I can’t listen to most satellite radio. It’s compression sounds to me like all the announcers have sinusitis and talk through a mailing tube.


Koz is talking about bitrate when you re-save (export). Bitrate is kbps = kilobits per second. So, if you know the bitrate and the playing time, you can calculate* the number of bits or bytes in the file. And, since there are 8 bits in a byte and 60 seconds in a minute -

File Size in bytes = (bitrate in kbps x 1000 x Playing Time in minutes x 60) / 8

File Size in bytes = bitrate in kbps x Playing Time in minutes x 7500

File Size in Megabytes = (bitrate in kbps x Playing Time in minutes)/133

Higher compression = lower bitrate = smaller files = lower quality.

Lower compression = higher bitrate = bigger files = higher quality.

Recompressing at a higher bitrate does not increase quality… The damage has already been done, and (at least in theory) a 2nd compression (even at a higher bitrate) results in more quality loss.

The LAME MP3 encoder allows you to choose VBR (variable bit rate) and target a quality setting rather than a particular bitrate. Then, LAME chooses the required bitrate moment-by-moment. V0 is the “best” setting and will result in larger files (and higher average bitrates). V9 is the highest compression for the smallest files.

Note: I tried reducing the sampling rate…

The sample rate (kHz) can be used to calculate the size of uncompressed files -

File Size in bytes = Sample Rate in kHz x 1000 x (bit depth/8) x Number of Channels.


  • The calculations are approximations. They don’t include file headers, tags (artist, album information, etc.), or embedded images (album artwork).

Many thanks to both of you. I’ve been fiddling with the process, and I think I’ve found that the project rate is the export rate, no?

This is what I do: import the file, stereo to mono conversion, change tempo 25%, edit metadata, set project rate to 16000, export. The 16000 seems to provide as much clarity as I need; as I said, it’s all just podcasts that are heard once and deleted.

Occasionally, about 5% of the time, I get an exported file that is either extremely slow tempo or extremely fast, and I don’t really know why. At present I just test each one before loading it onto the player.

If anyone can give me a pointer or two about smoothing out the process, I’m all ears. And thanks again for the input.