Audacity 2.0.0 (apparently from audacity-win-2.0.exe)
Windows XP, SP3
I have 130 MP3 files of 30-60 min. each that I would like to batch process. In some prelim. tests, I found that a file was twice as big as the source. So I went back and made adjustments and now the exported file is the same: mono, 22050Hz, 32 bit float. But the exported file is still twice as big. I repeated the same process with a small file, doing nothing to it but importing it and exporting to a new name. That file is four times as big as the original while the track info is the same: mono, 11025Hz, 32 bit float. I altered the sample rate to 16 bit but got no change in file size. Changing the default sample rate doubled the file size yet again. Clearly I’m doing something wrong.
Ultimately, my goal is to normalize these files which come from many sources with wildly different recording volumes. How can I ensure that Audacity produces a file that is little changed except for the normalizing and, in particular, that I don’t end up with many files that are much bigger than the originals?
MP3 is a compressed data format. To change the size you can change the amount of compression that is used when encoding. Greater compression produces a lower bit rate (kbps) which produces a small file. The downside is that the more compression, the worse the sound quality. MP3 compression always does some damage to the sound quality and the damage in permanent and cannot be repaired.
Processing an MP3 requires that Audacity decodes the MP3 into normal (uncompressed) audio data before processing and (assuming that you want the result in MP3 format) re-encodes the processed audio when you export. The re-encoding adds additional damage to the data. Each time the data is encoding, more damage is added and the more compression that is used, the more significant this damage is.
To set the bit rate (kbps) of an exported MP3, click the “options” button in the Export dialogue screen (the screen where you select MP3 and enter the file name).
Depending on your media player, you may not need to process the audio at all.
Many media players support “Replay Gain” or something similar (Apple use their own version which is called “Sound Check”).
How this works is that you use your media player (or other program) to scan the files and it will work out how loud it is. This information is stored in the file, so (if enabled in your media player) the playback volume will be automatically adjusted to suit the loudness of the file. This method is preferable (if your media player supports it) as there is no loss of sound quality (also it is quick).
Thanks so much for “replay gain”; I didn’t know about that. I’m sure the target player does not do it, but it’s good to know.
I’m aware generally of data degradation, but thanks for clarifying that Audacity is kind of starting from scratch and re-compressing. Bit rate is the key to file size, then, as I understand you. I’ll just have to experiment and optimize. Happily, these are all spoken–books on tape, sort of, and ancient radio broadcasts, but no music. So I may have sufficient room to play here. (Aha! I did dub some of this myself. So I should go back to the original *.aup on those! Thanks for drawing my attention to the process.)
I’m going to want to set up a chain ultimately. Bit rate is controlled in the Export dialog’s Options and the setting will hold until changed? All other settings will follow the original file (sample rate, sample format)? Anything else I should be paying attention to?
For speech, ensure that all of the files are mono before you encode (no point in “wasting bits” on making a stereo recording).
Try the preset called “Medium”. For mono voice that should give a good amount of compression and quite good quality. If you need a smaller file than that you will need to go to the low CBR settings, but then the sound quality is likely to deteriorate quite noticeably.
Thank you so very much. I’ll have to assess just how much growth I can tolerate but, as it turns out, quality differences are hard to hear with these files, even at very low bit rates. I so appreciate your pointing me in the right direction.
I have another question, but I’ll give it a new subject line.
