Hey there, so I’ve got an issue (That seems to persist with a bunch of people), where my project audio is at a perfect volume, and when I export it to 16-Bit Signed PCM WAV for a program called SecondLife, the audio is WAY too quiet (For some strange reason), BESIDES the volume slider for Master and Sound being at halfway (I guess spatial audio is just really screwy), and since WAV files are NOTORIOUS for clipping… could someone explain, in Goku terms, how to counter this? Since I did some digging and saw other people go NO proper results…
Or am I just right screwed and have to deal with the crappy build of spatial audio? Been having this issue for a while and was never able to ask
0dB is the “digital maximum”.* Regular (integer) WAV files, CDs, ADCs (analog-to-digital converters for recording), and DACs (digital-to-analog converters for playback) are all hard limited to 0dB and you’ll get clipping if you try to go over.
Audacity uses floating point for internal processing so for all practical purposes Audacity itself has no upper (or lower) limits and it won’t clip. If you export-as floating-point WAV that won’t clip either but you’ll clip your DAC if you play it at “full digital volume” and it’s “bad practice” to release a file that goes over 0dB.
With integer formats 0dB is simply the highest you can “count” with a given number of bits. For example, with 16-bits you can have values between -32,768 and +32,767 and if your negative & positive peaks hit those values that’s 0dB and you can’t go any higher. You can have higher numbers with 24-bit files but at playback time everything is automatically scaled to match your DAC so a 24-bit file isn’t louder than an 8-bit file.
The “catch” is… We’re talking about peaks and peaks don’t correlate very well with “loudness”. And, one peak in a long file limits how loud you can go linearly without clipping. Loudness is more related to the average (or RMS level) and loudness also depends on the frequency content, with your ears being most sensitive to middle-frequencies. That’s where limiting and compression can help. Compression and limiting “push down” the peaks or loud parts and then make-up gain can be used to bring-up the overall-average volume to make it louder.
Technically, the digital reference (and “maximum”) is 0dBFS = zero decibels full scale so digital dB levels are normally negative. With acoustic volume (dB SPL = sound pressure level) the 0dB reference is approximately the quietest sound that can be heard so SPL levels are positive. There is a direct correlation between digital and acoustic levels but no standard calibration since the acoustic loudness depends on your volume control, and your amplifier and speakers, and how close you are to the speakers.
So if I’m understanding this correctly, and don’t hesitate to correct me if I’m wrong, if I export it as a floating-point WAV rather than a 16-bit signed PCM, and THEN convert it to that, the audio reduction should counter the clipping? Apologies if this isn’t correct…
So if I’m understanding this correctly, and don’t hesitate to correct me if I’m wrong, if I export it as a floating-point WAV rather than a 16-bit signed PCM, and THEN convert it to that, the audio reduction should counter the clipping?
What audio reduction?
If you open the floating-point WAV in audacity and then run Amplify (with a negative amplification) or the Normalize effect, you can safely reduce the level.
Correction: This appears NOT to be the case, because once I mess with the gain, Audacity happily clips! Go figure, right?
Audacity “shows red” for potential clipping. Anything that goes over 0dB (or has something like 2 or 3 0dB samples in a row) will show red. It’s not looking at the wave shape. You can have a file that shows red and it might be clipped or it might go over 0dB without clipping.
On the other side of the coin, you can have a badly-clipped file and if you reduce the volume by a fraction of a dB it won’t show red, although it is still clipped.
When I upload the audio to SL, the volume gets reduced, and I’m trying to counter that.
And clipping is what I’d like to avoid while still having nicely leveled audio, what appears to me is happening is that whenever audio is used in a spatial environment, the way the falloff is designed to work is rather unrealistic, where it just sorta dips down to a point in a given distance, then slowly falls off in a slightly more natural point. I’m sorta lost here…
Has this ever worked? Audacity will cheerfully record live performances to perfect loudness, peaks, and background noise, assuming you started out with a nice, quiet, echo-free room. You can go from that straight into podcast or audiobook production with little or no extra compression or other volume manipulation.
What you can’t do is pull home recording work like that into competition with commercial music, theatrical, or broadcast productions, They’ve all been processed to death to make them artificially louder and more “marketable.”
Home Audacity users run into that all the time. “How come my guitar music is a lot quieter than the productions from Glen Glenn Sound.”
Audacity doesn’t clip internally. Because of the 32-floating internal format, if you apply an effect that causes the waveforms to go too high, they’re not gone. They’re up there waiting for you to bring them back down with another effect, filter, or setting. If you try that with a WAV file either before or after Audacity, everything that clipped is permanent trash.
Most production studios do not work in 32-bit floating. Studio productions can work in 24-bit, 96000 which is a regular sound format. Their job is to provide work to be available for production, reengineering, and mixing, later.
Are there formulas, manuals, or instructions you’re following for this production job? Reading through the postings, every time you touch the program or process you get unexpected or damaged results.
Audacity is doing what it’s supposed to. I’d be diving for the Spatial Sound Manual.