I was thinking of buying a Zoom F3 Field Recorder that allows for recording in 32 bit float.
In Shotcut, would I have the ability to raise or lower the levels if required, but still maintain the
low noise floor that 32 bit float is known for.
I would like to know how Shotcut handled 32 bit float audio.
I’m guessing 32-bit audio will work as expected.
I do not have a sample 32-bit WAV file from an F3 or F6 recorder. But I hacked up some 32-bit float files from my music library to simulate an F3 like this:
ffmpeg -i "03 Splanky.mp3" -t 00:00:30.000 -filter:a aformat=sample_fmts=flt,volume=50dB -c:a pcm_s24le -y splanky_s24le.wav
ffmpeg -i "03 Splanky.mp3" -t 00:00:30.000 -filter:a aformat=sample_fmts=flt,volume=50dB -c:a pcm_f32le -y splanky_f32le.wav
Both commands add 50 dB of gain to the music to drive them hard into distortion. The first line saves the output as classic 24-bit, which distorts terribly. The second line saves as 32-bit float, which also distorts on direct playback.
Then I bring both files into Shotcut and add Gain filters to them at -50 dB. The 24-bit file sounds nearly silent, as expected. The 32-bit file sounds like the original music file, at normal volume, as expected, no distortion.
I exported the -50 dB project using the 32-bit source, to a 24-bit output, and the export sounded like the original music file, as expected.
The only “problem” that I encountered was no audio waveforms visible on the timeline clips. But this is strictly a UI problem, not an audio processing problem. Actually, the waveforms might be there, but look like a flat zero-line since 32-bit has around 700 dB dynamic range and my music doesn’t even scratch it.
I would assume you’re good to go for an F3. Enjoy it!
Thank you very much for your test!
This means I will go ahead and buy the F3 when available, as these too are
low in supply, due to the chip shortage.
Again, thanks for all your help.
Sounds great! I did not test every audio filter. I have a hunch that some filters, perhaps the compressor and limiter, may want to convert to 24-bit in order to give context to “where is 0 dBFS when defining thresholds”. If this happens, use the Gain filter to bring the 32-bit audio into non-clipping range first, then proceed with whatever non-float-compatible filters you find after that, if any.
I would keep a close eye on the levels.
According to the Zoom F3 specs, it will accept up to +4dbU from my board,
so as long as I am not sending above this to the dual converters I will be good.
I can actually record a little lower, as I would could always use some subtle gain in Shotcut,
which is generally what I do with my present H1 recorder, or my Panasonic FZ-1000 camera.
(when I record to my camera, which is mic level I have to come out of subs 3 and 4 and
attenuate these faders down 30 to 40db, which is a suboptimal solution at best.
Regardless, if I come in hot from my camera or H1 they are both 24bit and can’t get rid of flat lined
distorted sine waves, for example.
As this is my synths coming from my board, I feel the F3 would give me some breathing space,
as once I hit record, my hands are full playing the keys and I can not ride the gain on the board or the H1.
As you showed me the 32bit works as expected in Shotcut I can now up my Youtube game with
higher quality sounding ShotCut videos.
Live music sounds like an excellent use case for float storage. From a fellow musician (saxophone player), good luck!
This topic was automatically closed after 90 days. New replies are no longer allowed.