After using the “amplify” feature to perfect the audio level (for a radio station), then exporting the song, the level changes randomly. When I reload the song into audacity after completing this process, then analyze the song again using “amplify”, there is still a change recommended even though I have JUST completed this process (I think the recommended change is always a different amount than it was the first time). It’s as if though the process of exporting from and/or loading the song into Audacity changes the level by some random amount. The files are mp3, and I am keeping the bitrate the same constant amount as it was to begin with. I’m using Windows 10, 64-bit, and the program version is 2.1.2 Any help with fixing this problem will be appreciated as I am beyond frustrated. Thanks.
MP3 is a “lossy compression” format. It is not exact.
Encoding to MP3 is a trade-off between file size and sound quality, which is in part accomplished by discarding some of the audio data, so there is always some loss of sound quality each time the data is encoded. An MP3 file is never identical to the original data that it was made from.
Thanks very much for the reply. Can you recommend a better quality format that still doesn’t take up too much storage space? If not, I will go on using mp3’s and just accept it’s faults and assume that they are already the best balance of quality and compact size that’s available given their popularity, especially since I know I can not improve the quality of something that was an mp3 to begin with.
I forgot to mention that, while I did know that the quality drops each time it’s encoded as an mp3, I didn’t realize that the volume level is a part of that drop. Thanks for clearing that up, I’m glad to know I’m not doing anything wrong and it’s just how it is for everyone.
Although the wave shape changes making some peaks higher and others lower, with lossy compression the average/RMS and loudness shouldn’t change.
Some people like to normalize for -1dB peaks (or so) before compressing to keep the MP3 from going over 0dB. But, the MP3 itself can go over 0dB without clipping so I don’t worry about it. (You can clip your DAC if you play-back the MP3 at full volume.)
If you wanted to, you could do a 2-step process where you export to MP3 to find out how much peak change you’re going to get, and then start-over with the original file, making the appropriate adjustment the 2nd time.
OGG files provide about the same (or slightly better) size / quality trade-off without changing the peak amplitude.
(OGG also keeps the length the same whereas MP3 adds a tiny bit of padding to the start).
The disadvantage of OGG is that it is less well known, so not everyone has the codecs installed, for example, neither Windows Media Player or iTunes support OGG by default, though most good audio players do support OGG, and “DirectShow Filters” are readily available that allow Windows Media Player, and other DirectShow-based players to play OGG files.
More information about OGG on Windows: http://www.vorbis.com/setup_windows/
In some ways I’m more confused than when I started. I’m starting out with mp3’s, and just trying to make sure the level is plenty loud without being at all over-modulated. I was in radio in the 90’s when you had to manually adjust the levels because computers didn’t do it for you. I want the loudest quick bursts of the loudest songs to maybe shoot briefly up to as high as +2, but not “camp out” there. My understanding is that if I adjust a song’s peak amplitude to the recommended 0db, then the VU meter would NEVER surpass 0, and that’s not what I want. Nearly every mp3 that I analyze recommends shaving off a little bit of sound. In my understanding (which I assume is wrong), if I want the loudest parts of the loudest songs to reach as high as, lets say mid-way between +1 and +2, then I should set the peak amplitude to +1.5 I have a strong feeling that this thinking is WRONG. Can anybody please explain to me, in fairly simple terms, why this thinking is (probably) wrong?
MP3 files can store values above 0 dB, but can the digital to analogue converter in your sound card send signals at +2 dB without distorting in your playback equipment? Probably not.
I was in radio in the 90’s when you had to manually adjust the levels because computers didn’t do it for you. I want the loudest quick bursts of the loudest songs to maybe shoot briefly up to as high as +2, but not “camp out” there.
The analog signals in the console (mixer) can go over 0dB. There is a limiter on the transmitter to prevent over-modulating, and I believe 0dB on the meters is 100% modulation modulating (and there is usually some additional compression somewhere in the chain).
The last time I was in a radio station (in the 1970s) it was common practice to go occasionally “into the red”.
if I want the loudest parts of the loudest songs to reach as high as, lets say mid-way between +1 and +2, then I should set the peak amplitude to +1.5 I have a strong feeling that this thinking is WRONG.
Yeah, I’d say it’s “wrong” because your DAC is hard-limited to 0dB and if you play the file with the digital volume control set to 100% your DAC will clip. “Regular” (integer) WAV files and CDs are also hard-limited to 0dB.
Technically, there’s no reason to clip or to go over 0dB because digital has a wide dynamic range. Back in the analog days you needed to keep a “hot” signal to overcome the tape noise. With digital, that’s just not an issue. …If you want it louder, you can turn-up the amplifier playback volume.
But personally, I don’t worry about MP3s going over 0dB as long as the uncompressed original didn’t go over 0dB. I’ve never heard any distortion caused by that (I have heard bad-sound low-bitrate MP3s), and if you do hear distortion you can always turn it down. And, I use ReplayGain (or MP3gain) which ends-up turning-down most songs anyway…