I understand that Audacity may have a statistics tool related to the lowest and highest audio volume. If not, there may be a plugin that performs this statistic, if it’s digital, even better. That is, to measure the audio in Hz of the two audio peaks (lesser and greater).
If so, any tips, guidance or teaching will be very welcome.
I always normalize my audio at -3 dB however I have carried out a test in a plugin called Relay to make this measurement however, even though the audio is at this normalization of -3dB this plugin like the others for example at certain times of the audio there are quick peaks exceeding 0 dB. Is this normal for a spectrum analyzer or even this plugin I mentioned above?:
If this should not happen at any time, what would be the most practical and viable solution?
Regular normalization (peak normalization) is pretty-much foolproof. If you normalize to -3dB in Audacity the highest actual peak(s) will be at -3dB.
Some analysis software looks for “inter-sample peaks” and that can show higher than the actual digital data. There is no actual inter-sample data in the digital domain, there are only samples. The inter-sample peaks are an estimation of the reconstructed continuous analog output from the DAC (where there are no samples).
With regular program material there are thousands or millions of peaks. Every wave cycle has a positive and negative peak, and two zero-crossings. And there are an unlimited number of frequencies, with many frequencies at the same time.
Spectrum analysis is complicated and imperfect (1) But it the peaks shouldn’t exceed the real peaks. And usually the spectrum analyzer peaks are lower because the real peak is the sum of the frequencies occoring at the same time.
(1) Theoretically, it can be “perfect” if you have the same-continuous sound for an infinite length of time. With real audio changing moment-to-moment you have to make compromises. And FFT (used for digital frequency analysis) divides the spectrum into frequency bands so you don’t see every individual frequency.
Thanks for the teaching, I confess that I don’t have a deep knowledge of audio like you, but I understood some of the nuances that you explained.
I also used an analyzer called SPAN, whose image is attached. The same thing happens, even though the audio is -3 dB normalized in Audacity, at times the signal is saturated by observing the led on the ruler.
So by your explanations this would be normal, or would I have to decrease the output gain to avoid this saturation even just in a few moments?
Continuing my desire to learn, how will I know the difference between the real peak (lower and higher) so that I can use the compressor or the limiter for example?
Does Audacity have this tool? If so, which way to get there?
Thanks for helping me again, you guys are very knowledgeable about the process.
Sorry but could you teach me in more detail your explanation? For example I don’t understand: "Since the built-in volume is what your platform (YouTube etc) requires, so do you reduce any loud (+dB) peaks with Audacity’s (soft) limiter? As I said, I’m a layman in comparison to your knowledge.
For example: I don’t know what it is: LU, LUFS, LKFS, LRA and PLR measurements.?
But I downloaded the software and installed it, it is working normally as per the attached image.
I will read the manual carefully anyway, any teaching and guidance will be very welcome.
The example you’ve shown has “integrated loudness” of -16.8 LUFS.
Different platforms have different loudness standards,
( e.g. on YouTube the maximum loudness is -14LUFS).
Your example has a “true-peak maximum” of +0.4dB, (i.e. clipping a little bit, probably inter-sample).
Hello Trebor, good morning (In Brazil)! Really this tool is exceptional for audio analysis. I have a lot to thank you for nominating me, without you I don’t think I would have ever known her. In the case of the LUFS unit, the greater the numerical indication, the less audio volume? Friendly hug!
