Just got a new plug-in and now I can get my RMS way up (or down) – in other words, louder. Previously I was lucky if I could get as loud as -20 RMS-A (RMS Average using the wave stat analyzer, stats.ny plug-in). So now I finished a song at about -17 RMS-A and it was pretty loud compared to others online – so what is a good RMS A-average value? A value to fit in with other songs, as it seems most are a similar value /loudness. Thanks…Ron.
I did a Googl search and found this, here, on this site, from 2013:
Some of the technical detail is a little bit “out”, for example:
“There should be exactly 500ms (0.5 seconds) at the head of each file”
The start time of an MP3 file is not “exact” due to limitations of the format - what they mean here is “There should be very close to 500ms (0.5 seconds) at the head of each file”.
“Your submitted files should measure between -23dB and -18dB RMS”
This specification is incomplete as they do not state the RMS “weighting” or “window size”. In practice the exact definition probably does not matter too much - I assume that they are just trying to give an indication that the recording should be “reasonably” loud without pushing the loudness too much - the compression / limiting steps that they describe are likely to be close enough to the right ball park figure. Comparing your recording with some of their published recordings will immediately show you if your recordings are too loud or too quiet. There are also some plug-ins that can give you an RMS and peak amplitude figure for your recording such as the “Stats” plug-in available here: viewtopic.php?p=99454#p99454
So I guess this answers my question. Although as far as
“Your submitted files should measure between -23dB and -18dB RMS”
Yes I too wonder if thats the first value RMS or the second “weighted…” given by the Audacity wave stats analyzer?
That was referring to voice recordings for audiobooks, not for music.
A-Weighted RMS was intended for measuring low level noise, such as the background hiss created by electronic components in audio circuits. It’s not a particularly good match for assessing “loudness” for music at normal listening levels. C-weighted or k-weighted would be better for that (though I know they are not yet available in the “stats” tool).
For music, there really aren’t any set standards, until it gets to the radio/TV station (then it gets hit by the broadcast standards).
There has been a strong trend over the last 50 years or so for commercial recordings to be made increasingly loud. (see http://en.wikipedia.org/wiki/Loudness_war). Happily there seems to now be elements of rebellion as “dynamic range” shows signs of becoming fashionable again, at least with some genres of music.
Thanks Steve, but I seem to get good use from the wave stat tool (stats.ny), with music (or talk). But will look at the RMS value now as opposed to the RMS weighted value as I was before, to set some kind of standard, if only for my own songs, to keep them of similar loudness.
In an research, I shifted down all three of my files and decreased all of their RMS principles from around -20dBFS to -16dBFS, using the ‘Gain’ device in ProTools. In doing this the waveforms efficiently had an RMS value of -16dBFS however, they all then had a optimum outcome obtain of 0dBFS. Looking at the waveforms it seems as if some areas had connected, but there is no distortions obvious in the site files. Does anyone know what is occurring here? I have connected an image to this line.
–Not sure what level of a ‘producer’ you are but just in case you didn’t know : Decreasing an RMS value like from -20 to -16 is increasing the volume (i.e., loudness, gain, decibels). And going as high as you describe (visually) is probably clipping. Audacity has an option to show clipping in red lines or you can turn it off. If you’re viewing in Audacity you may have the “show clipping” option off. I don’t know, just a guess. I personally try to keep things (amplification wise) no higher than -0.1 dB from clipping --although some say you should keep it at least -1dB below–I’m not clear on that, and with my poor recording equipment, I need all the volume I can get. I’ve examined professionally done music (albeit ripped from my CD to a wave file) that was at 0dB’s. [Above zero decibels is clipping]
Found this plug-in (through Audacity Downloads-Plug-ins): “DFX RMS Buddy v1.1” --so far much better than stats.ny for measuring RMS. Just started tinkering with it so will see… but to use you just highlight a track, regardless of length! And then open it from the effects menu and click “OK” and it takes a few seconds doing it’s thing and then it exits. Now pull it back up/out from the effects menu and it has the reading on it. Kind of awkward but simple and hopefully acurate–Seems to be so far. Here is the link: http://www.hitsquad.com/smm/programs/DFXRMSBuddy/
It’s also interesting to view professionally done songs and their varying values.
Is that just because it doesn’t have the 30 second limitation, or are there other advantages?
Do the RMS measurements agree with stats.ny?
I haven’t thought to test them on equal par --per a same 30 second sample. Will do ASAP. But yes, I like the ability to check the whole song as opposed to being limited by a 30 second sample maximum. The only other feature I’ve noticed the RMSBuddy has is it gives the peak readout as well, how many decibels under 0. There’s also a window with a “LOAD” and “SAVE,” that I’m not sure what they do. Just thought what that might be…maybe it saves a particular RMS so you can apply it without having to do the math to come up with the correct “Amplification” effect setting needed to set a particular RMS value. That would be great, I guess that would make it an RMS normalizer as well. I failed to read any description that came with this plug-in describing what the LOAD feature’s for…will report back ASAP.
UPDATE: I’ve not been able to get any functionality out of the RMS Buddy’s “Presets,” “LOAD” or “SAVE” features and have no idea what they are for! When I downloaded this plug-in effect it came with a file about copy rights and a file showing a small brief description of it’s functionality that never mentions the presets, load, and save features.
Please do, I’d be very interested.
A bit of background: I wrote stats.ny for a specific project that only required short term analysis, and it was easiest (and quickest) to write it for short term analysis only. Since then, there have been a few requests for longer analysis, so I’ve written that (…somewhere). I’ll be busy with gigs over the next couple of days, but will hunt it down and post it next week (please post me a reminder if I forget )
The new version also has C-weighting.
I thought that was in stats.ny. If not, it’s easy to add.
There’s an 1-hour version already in the wave-stats topic (one of the last posts).
Overall peak, positive and negative as well are already integrated.
Rms normalization is easy:
If the plug-in shows -27 and you want -20 dB FS, amplify by 7 dB.
You have probably to check “Allow Cliping”.
Use a Limiter to bring the peak back to under zero.
NEXT POST (scroll down)
GREAT!! --and “C-weighted” may be the be all end all–I hope. Because I have discrepancies between TALK and MUSIC. Talk (with silence in between) resulting in a lower RMS value then it should be - I’ve explained this problem better and will post NEXT as well (on this thread).
Yes, Wave STATS… does show peak values as well, and more: DC off-set, etc. It slipped my mind at that writing. So, if Wave Stats… did not have a sample limit, I would consider it the better choice. And hopefully even better with the “C-weighted Average”
25 SECOND SOUND SAMPLE - ANALYZERS’ COMPARISON TEST
(In Audacity 2.0.5) [25 seconds chosen because unlike RMS Buddy, Wave Stats is AS OF THIS WRITING limited to 30 seconds; All figures are decibels]:
Analyzer 1: “Wave Stats…” (stats.ny, an add on plug-in that loads to the Analyze menu). Results: Peak level -3.5; RMS -21.7/RMS (A-weighted) -24.2
Analyzer 2: “Destroy FX: RMS Buddy…” (dfx RMS Buddy.dll, an add on plug-in that loads to the Effects menu). Results: Peak level -3.47; RMS -21.72/RMS (A-weighted) Not Available
Conclusion: With only one test (more may be required) they seem to be in agreement as to what the RMS value of this sample is. With the RMS Buddy showing more of a precise measurement*, while the Wave Stats seems to ‘round it off.’ The Wave Stats is more user friendly though, because with the RMS Buddy there seemed no way to get the reading in one try. It involved first: selecting OK (then it analyzes and exits), then I had to select it again from the Effects menu to view the results.
- Audacity rounds amplication figures applied as well (* Or maybe just in the readouts, not in the actual result [?]). For example, I amplified a song -2.51, Audacity showed -2.5 when I reviewed afterwards with the Amplification effect; and RMS Buddy showed it as -2.53.
So on another peak value test, I amplified a sample to -3.0 (avoiding any increments or fractions of a decibel, like -3.1 etc.) and Audacity showed it to be -3.0 but RMS Buddy showed it to be -2.97. I would conclude they are not exactly in agreement but only off by a small degree, 0.2 to 0.3 --with either one or neither one being exactly correct [?]. Close enough for me anyway.
Conjecture: If Audacity is off, and things are actually louder than Audacity shows. then finishing a track too close to clipping may actually be clipping. That may be another reason it is recommended to finish no louder than -1.0 dB’s.
All editing, testing, etc. performed with Audacity only.
More about the RMS Buddy add-on plug-in effect: I’ve not been able to get any functionality out of it’s “Presets,” “LOAD” or “SAVE” features and have no idea what they are for! When I downloaded this effect it came with a file about copy rights and a file showing a small brief description of it’s functionality that never mentions the presets, load, and save features.
I don’t understand – are you saying there is an ‘up to a 1 hour sample’ RMS analyzer plug-in available? Here, somewhere? Free? I searched “wave-stats” and did not see anything… If there is such a plug-in, a link would be appreciated, thanks.
I don’t think that 1/100 dB values make any sense (for the purpose of reading out measurements)
The measurements are rounded off from a 32-bit float number, there’s any precision you like.
The pauses in voice recordings shouldn’t alter the RMS value too much, i.e. push them down.
However, you can first apply “Truncate Silence” to bring the phrases together in a whole block.
Not from my experience… They were definitely resulting in lower RMS values [Like -25 as opposed to guitar vocals -15–and I know they’re off because if I adjusted them to be the same (RMS), the talk tracks would be much louder than the voc/guitar] --with stats.ny or RMS Buddy plug-ins
Wow-interesting–that would solve that --but I’ve now gave up on the whole RMS concept!! I think my ears and brain do a better evaluation [!].
Thanks for the link Robert J. H.
Here’s my current thoughts on RMS (I wrote earlier today) Thanks for sharing, everyone :
TRIAL BY EAR: aug4 CURRENT CONCLUSION: (nix past RMS entries of mine here/well, it is a learning curve–always/right?): calibrating volumes manually based on a standard/a snippet of Gasland (track 1 of my CD) /love the sound i got there (recordings w/an iPod mind you =( /so using that to adjust the volumes of all the others/by ear!! as no rms analyzer has proven as accurate! /nor the WMP’s automatic volume level option/when cd burning. -this was mentioned/the by ear best by someone here–unless weighted B/C?? or something comes to my usuage I have denounced/86ed RMS or RMS A-weighted as any use to me. -thanks for your time/…‘trial and error’.
UPDATE: burned a cd then listened in home stereo–close but still a few songs too loud (just a little)//–this is why I wanted a device to calibrate–but there may not be such a thing–such a thing as the combination of the human ear and brains ability to almost create a third sense–I wonder how the pro’s do it?? jeez.
I mean voice in the sense of narrations, not vocals.
They are definetly another thing.
Perhaps, it would be worthwhile to include an option “Ignore Gaps” over x secons and y dB.
RMS update. I think I’m starting to learn something about RMS–and a 10 sec sample is really all it takes to do some evening out.
So far in my current trial, if the RMS is the same, and usually first 10 seconds is all that’s needed to check-(with an RMS analyzer), unless, quiet intro or something. So with that, if the RMS is made the same for two songs (by plus or minus gain/amplitude) and they don’t sound the same volume it may be because one has more low-end noise than the other–requiring EQ to get the bass/mid/treble sounding similar, which then simultaneously brings the RMS value in-line.
UPDATE: oct13/2014 I have no frickin idea about controlling rms etc-- I still have problems with RMS (as any kind of guide) and am currently just equaling volumes of my different songs by using MY EARS. I then use RMS analyzer just to label it’s RMS value for reference sake (if only for that one song) --and maybe will see a pattern one day…[?]
Well hello (echo) --to update The Quest For The Perfect RMS Value, found this interesting link on the subject: http://homerecording.com/bbs/general-discussions/mastering/default-rms-level-songs-357180/
Interesting thread. Couldn’t figure out what they determined or was the most helpful.
I learned the hard way on my remix CD music venture I posted two threads on that RMS readings don’t foretell how loud it will be on headphones vs speakers only if reverb is added to mastered stereo music that has had reverb already applied. The RMS readings from Wave Stats said the data was VERY loud and it was on my headphones but not on my speakers. From my reading on the binaural effect of headphones and reproduction of reverb effects this audio phase inversion trickery is quite a rare thing to happen I’m sure and wouldn’t apply to your mixing recorded music but provided it as an example of what to look out for concerning relying on measuring.
I would suggest you keep tabs on highly directional frequencies around 200-10kHz which create most of the perceived loudness according to the A-weighted curve graph and sine sweep sound test on this page…
Note the frequencies that hover around 0db for the A-weighted curve on that graph which are considered the loudest. If you belt it out “yell” singing like Ethel Merman, there’s your sound check for loudness.
I acquired some understanding of audibly gauging overall loudness from a YouTube interview of old school mixing engineer Mick Gauzaski by listening to his Daft Punk “Get Lucky” mix which builds the composite tracks of each instrument & vocal arranged in a tightly clustered shape listening on headphones. The closer all the instruments sound arranged around center the better at judging overall loudness so all can be heard but one doesn’t overpower the other. All the balanced elements fit together like a puzzle. If it’s just you and the guitar, listen to the composite and check if you can switch back and forth within your headphones and just hear the detail in the vocal and then switch to just listening to guitar detail. One doesn’t drown out the other. When it’s balanced like that is when you make it louder.
You can hear this closely clustered balanced sound stage shaping on any pop style finished piece of music on YouTube. Note most pop songs have a limited number of tracked vocal & instruments where the more instruments the lower the volume for each so they don’t drown out the lead instrument or vocal. Not every instrument can be loud.
There’s even a recording of a symphony orchestra using only mic-ing techniques and no mixing or processing of any kind that adopts this clustered composite style soundstaging by this audio engineer. https://www.youtube.com/watch?v=eudIlbfpRpQ
This is the best recording of Tchaikovsky’s “Capriccio Italien” I’ve ever heard.
Every instrument is center clustered and upfront. You don’t need to turn it up to hear everything. I should know I played this piece as a bass trombonist in my University’s orchestra back in '78.