19dB is a lot of gain, and I’m not sure why your signal is so weak. A line-level RCA connection from a cassette deck, or a headphone connection with the volume turned-up should be stronger.
How has Audacity found the level 19.1 in my case? Is that recommendation of the software?
It scans your file to find the current peak. It means your peak was -19.1dB before applying amplification. If you apply that change and run Amplify again, it will default to 0dB of change, which is no change.
Many thanks for your patience! Can I see somewhere how the peak amplitude changes when adjusting Amplification (dB) slider?
Hang on, it get’s complicated…
dB (decibels) are a relative, logarithmic, measure. In the “digital world”, the reference is 0dBFS (zero decibels full scale). In the computer, 0dB is essentially the maximum count you get with a given number of bits. That means that 0dB is a bigger number with 16-bit audio than with 8-bit audio. But the audio software takes care of scaling, so a 0dB 16-bit file is not louder than a 0dB 8-bit file. (You don’t need to worry about the underlying binary numbers unless you are a programmer.)
Since 0dB is the “digital maximum”, we are usually talking about negative dB levels.
Floating point files use a different numerical reference, there is essentially no upper limit and they can go positive. Audacity uses floating point internally so it can go over 0dB. But, if you go over 0dB and you export to a normal (integer) wave file, the waves will be clipped (they’ll have distorted-flat tops & bottoms).
When we adjust the volume for 0dB peaks, we call this “normalizing”. It’s fairly common practice to normalize (maximize) digital files. Most commercial CDs are normalized, or at least the song with the highest peak is normalized.
With acoustic sound, decibels are measured in dBSPL (Sound Pressure Level) and here the 0dB reference is the smallest sound humans can hear. So, acoustic sound levels are positive numbers. Although digital files and acoustic levels are measured with different dB references, they still correlate. If you are listening to your speakers at 90dB (SPL) and you boost the volume by 6dB in Audacity, the sound level will go up to 96dB. But, since every playback system is different and every system has a volume control we usually don’t know the exact relationship between the digital level and the acoustic sound level.
A decibel number can represent a relative level or a change… Your file is peaking at -19.1dB, so if you amplify it by +6dB it will peak at -13.1dB. So, although we are mostly measuring negative dB values, a change can be positive (boost or gain) or negative (reduction or attenuation).
BTW - If the peaks are boosted by 6dB, the average level is also boosted by 6dB… All of the samples in the digital file are increased by 6dB.
A change of 6dB is a factor of two. That is, +6dB is twice the signal level and -6dB is half the signal level. In a digital file, that’s double or half the digital value, and with an electrical signal it’s twice or half the voltage. 20dB is a factor of 10.
Just to further complicate things… Power (Watts) is related to the square of the voltage. So +6dB is 4 times the power, +3dB is twice the power, and +20dB is 100 times the power.
Also, the peak level does not correlate well with perceived loudness. Our ears don’t respond instantly and many quiet-sounding songs have short-term 0dB peaks. Loudness is more related to the average level and the frequency content… If you normalize all of your songs, they will not be equally loud.