I just started using the new Zoom H1essential to make field recordings of our Big Band.
The recorder produces 32-bit float WAV files, which seem to import just fine.
With my previous Zoom H1, I would import the 24-bit WAV files, normalize the whole project, label the songs, and “export multiple” all the mp3’s. Since I always set my recording level conservatively to avoid clipping, normalizing usually brings the levels up.
With the 32-bit float input file format, the levels appear pretty saturated in Audacity already. Now normalizing brings the levels displayed in Audacity down, not up. Should I be normalizing at all, or just export the mp3’s without that step?
32-bit float has the ability to exceed 0 dB, but sound cards clip at 0 dB. Also, some MP3 encoders clip at 0 dB (I’m not sure if Audacity’s MP3 encoder does or not, but if not then the sound card will clip at 0 dB).
To avoid clipping on playback, the audio must not exceed 0 dB.
Audacity doesn’t show over 0dB so you won’t see the wave peaks. And Audacity will “show red” if you have Show Clipping turned-on but it’s just looking at the levels and showing potential clipping.
The 32-bit floating-point digital format has virtually unlimited range (almost -1000dB to +1000dB), but at some (unknown) point the Zoom will hit internal voltage limits and it WILL clip. So you should still TRY to keep it below 0dB, knowing that you have some headroom if you accidently go over.
Or, it’s OK to record lower. (Pros recording at 24-bits often record around -12 to -18dB.)