Proof of a theory that I thought was true anyway...

I had a fancy that a “recording” of nothing would take up the same space as a recording with an active audio track. It does, I recorded an hour of both and compared the side of the two and they were almost exactly the same.

Again I figured to to be true but ran the test anyway. Use of Timer Recording helped to make the two files the same size.

From by recording days and using tape, One hour of tape is one hour of tape with or without sound track…

I believe being sheltered in place had something to do with this little ditty…

Absolutely right :slight_smile:

A bit more detail:
If you are recording with Audacity’s default settings, the sample rate “44100” means that there are 44100 samples every second. Audacity’s default format is “32-bit float”, which means that every sample has 32 bits (binary digits), which is 4 bytes (8 bits = 1 byte).
Therefore there are 8 x 44100 = 352800 bytes = 352.8 kB per audio channel for each second.

If the recorded track is silent, then each of those 32-bit numbers will have values close to zero. For normal “sound”, the samples vary between positive and negative values and should be in the range +1 to -1. These are the values that you see on the vertical scale of an Audacity track.

There is an interesting exception to the above. If you “generate” silence (Generate menu), then Audacity knows that every sample will have a value of exactly zero, so it does not bother creating the actual samples. It just keeps a note in the AUP file of there being an imaginary block of zeros in that part of the track. This kind of “imaginary” block of zeros is called a “silentblockfile”. When you export the silentblockfile, or if it included when processing audio, it gets converted into actual audio samples. This is why it is possible to generate very long silent tracks almost instantly - very little data has to be written, just one entry in the .aup file for every 262144 samples and no data.

You say “nothing” but odds-are there would be some low-level noise, not true flat-line silence.
If you generate true flat-line silence there would be a big difference in size if you exported copies in WAV & FLAC.

Many compression methods are “smart”.

I just did an experiment - I exported a song as MP3 with Variable Bitrate and “best quality”. I Silenced the file with Generate Silence to get a silent file with the same playing time. Then, I exported the silence with the same settings. The silent file was about 1/8th size of the music file.