Understanding sound level

win 7 audacity 2.0.5

this is my first post, so hello everybody - I am very new to Audacity and have a very limited understanding of recording sound, so please feel free to explain things from the very basics, I am really struggling with some of this stuff, particularly the terminology.

I am wishing to record some music from the computer and convert it to an MP3. I have so far been successful apart from the end product is quieter than other MP3 files. Firstly when recording the input level is maxing at about "-6"ish. The resulting visual of the Audio track seems to peak at about 0.4.

Using “Normalize” from the effects menu seems to increase the peaks to about 0.9 but the resulting MP3 is still a little quieter than other MP3s - I am not understanding what the other settings are within the Normalize dialog box?

Using “Amplfy” seems to suggest a level that takes the peak level up to about 1 (very similar to Normalize) and if I increase this level it warns of clipping (I presume this causes distortion when the “1” barrier is exceeded.

I still tend to think it is still a little quieter than what I am aiming for - any suggestion ?

1.0 or 100% is 0dBFS (zero decibels full scale). Your ADC (analog-to-digital converter), DAC (digital-to-analog converter), and most file formats are hard limited to 0dB. That’s considered the digital maximum and if you try to go over you can get clipping [u]clipping[/u] (distorted flat-topped waves).

Audacity itself can go over 0dB without clipping (although the audio may clip if it goes over 0dB and you send it full-volume to your soundcard’s DAC).

MP3s can go over 0dB. In fact, since MP3 is lossy and it changes the wave shape, some peaks get higher and other peaks get lower and it’s not unusual for a song that’s normalized for 0dB peaks before compression to song to have peaks of about +1dB after MP3 compression. (But, this doesn’t make it “louder”… See below.)

Normalization sets the peak at exactly 0dB (or whatever level you choose, but 0dB, or very close to 0dB is standard). This is as loud as you can go linearly (without distortion or an effect like dynamic compression or limiting). GoldWave (another audio editor) calls it “Maximizing”, but “Normalizing” is the more common terminology.

Now, here’s the catch - Your ear doesn’t respond instantly to peaks and a short-term 0dB peak won’t sound loud… Loudness is more related to the average level than to the peak. It’s also related to frequency content… Our ears are more sensitive to mid-frequency sounds (around 2kHz) than to deep bass or high frequencies. A lot of quiet sounding songs are normalized with 0dB peaks. Most commercial recordings are normalized, but obviously there are quiet songs and loud songs.

Most commercial recordings have LOTs of dynamic compression (totally unrelated to file compression like MP3 compression). Dynamic compression (and limiting) can be used to boost the quiet and “average volume” parts of the song without boosting or clipping the peaks. This brings up the overall loudness of the recording. But of course, this reduces the dynamics of the music and when it’s constantly-loud it can become boring.

If you want to match the volume of all of your recordings, you can use [u]MP3Gain[/u] or one of the ReplayGain variations. Or, if you have iTunes or an Apple device, turn on Sound Check. These tools analyze and match the loudness rather than setting the peak level (or average level). However, since many quiet sounding songs can’t be boosted without clipping, these tools match volumes by reducing loud songs which often ends-up making most of your music library quieter.

Firstly when recording the input level is maxing at about "-6"ish. The resulting visual of the Audio track seems to peak at about 0.4.

-6dB is half.* So, the biggest peak should hit 0.5, but you may not see the peak unless you can find the spot and zoom-in.

Note that dB is relative so we can say, “Set the peaks to -6dB” (or more properly -6dBFS) which means the peaks are 6dB below the 0dBFS reference. And if we then reduce the level by 6dB (relative to the current level) the peaks will be at -12dBFS.

When measuring acoustic loudness (loudness of sound in the air), the 0dB reference is at the opposite end. 0dB SPL (Sound Pressure Level) is the quietest sound we can hear. So, SPL loudness levels are measured in positive dB. i.e. A loud rock band might be 100dB SPL.


  • A 6dB change is a signal level factor of two. i.e. +6dB is twice the value in a digital audio file or twice the voltage. But, 3dB is twice the power, and 6dB is 4 times the power. If you’ve got an amplifier putting-out 25 Watts and you boost the level by 6dB, it will be putting-out 100 Watts.