Win10 Audacity 3.1.3
I record an audio track (sung or spoken) and a 3rd-party program reports back that “The volume falls outside of the target range (86–92 dB).”
How, in Audacity do I
(a) Specify that the volume level is to end up between 86 and 92 db and then
(a) What Audacity command takes that setting and applies it.
It seems to me that this ought to be a simple operation, but I can’t find out any reference to it.
I have used the Effects, Amplify before, but the setting there seems to say “reduce it by this much” or “increase it by this much”, so I use a time-expensive trial-and-error method:-
(1) apply -2.0 or +2.0
(2) re-save as MP3
(3) send through the program again
(4) again get told “The volume falls outside of the target range (86–92 dB).”
(5) and still don’t know whether I am falling off the high end or falling off the low end …
Hi Trebor, I have read the LibriVox manuals rather fully. The page you mention says “Volume: around 89 dB (-16 to -21 LUFS)” which is less definitive than the output of Checker.exe as quoted in my original post, and then does not explain HOW to coerce the volume to that specific figure (or in Checker’s case) to that range.
Be aware that, by my understanding, Checker.exe was written by the spouse of a LibriVox member who is no longer around, so Checker.exe is unsupported by source code or by documentation, let alone by programmer. Checker.exe is a take-it-or-leave-it black box.
I tried "normalizing to "; in Audacity, Effects, Loudness Normalization… using the default -23. The wave form changes, perhaps for the better, perhaps not, but still this does not tell me that the volume is now in the target range (86–92 dB).
The message from LibriVox’s pet utility Checker.exe tells me that the volume is OUTSIDE the range; so the question is “How do I tell Audacity to make sure that the volume is within the range 86–92 dB”?
It may be a fault of Checker.exe that it expresses its displeasure in dB units “86-92”, and perhaps the programmer should have expressed it in LUFS units. In which case my question translates into “How do I translate from dB s to LUFS?”
I note, too, that just like Amplify, the Loudness Normalization allows me to enter a scalar value which appears to reduce the height of the wave form, making the track quieter, for sure, but still does not allow me to quieten it to a specified value.
My situation makes me look for some sort of Audacity command or macro that says “transform this audio track so that it falls within the range 86-92dB”
Thanks, Chris
Thank you Trebor.
I now have an MP3 file that passes the Checker.exe program.
I chose -18 LUFS as being a middle-of-the-road value, but of course if this set is a set of exponential values …
(a) That checker displays an error only in terms of a range of dB diverts my attention from LUFS units.
(b) I now have a means (Effects, Loudness normalization) of coercing an MP3 into a range acceptable for Checker
(c) Excepting trial and error (21, 20, 19, 18, …) I see no way for the user to make a simple command/request to lower the volume to the top (or perhaps the bottom) of Librovox’s Checker.exe specified range.
Bottom line (for me) if Checker.exe had expressed its petulance in units of LUFS as well as dB, I might have gone looking in the Librivox and Audacity documentation for LUFS, so I think that the outdated Checker.exe is a cause of mis-direction here.
Thank you, again, for your patience and encouragement.
Chris
Librivox should also specify the maximum peak acceptable, (I can’t see one in their specifications).
After loudness normalization, I would apply a (soft) limiter -2dB, without make-up gain,
i.e. ensure headroom of 2dB before making the mp3.
[ as the conversion to mp3 can increase the peaks by about 1db.
Without headroom the mp3 could be clipped ].
Hi again, Trebor.
Librivox is a bit sloppy in some of its specifications/rules/instructions, it is true.
… I would apply a > (soft) limiter > -2dB, > without > make-up gain, i.e. ensure > headroom > of 2dB before making the mp3. [ as the conversion to mp3 can increase the peaks by about 1db. Without headroom the mp3 could be > clipped > ].
Filed for future reference.
My experience with Librivox is that mnaking something the-best-that-it-can-be is often overkill. There is an adoption of weak standards, disclaimed by “You must remember that we are all volunteers here …”
I can feel an enhanced Audacity macro creeping up on me, aimed specifically at conversion to MP3
