How loud is 0dB, and what is a good dB to mix to?

Question

In Shaun's recent post about the sound design challenge, he mentioned that there was a request for a loudness-limit (-10dB peak, I believe), and I realized that I was most likely the reason for it, if only in part. See, I've been mixing to 0dB average -- because I really know no better -- and I didn't know until now that it was even a wrong thing to be doing.

In none of the books that I've read have they mentioned a general dB to mix to, at least not that I can remember (it's possible it's mentioned, but if it is, it's mentioned only once or twice and not with much weight).

So... how loud IS 0dB? Coming from a guitar background, I assumed 0dB was "unity gain," and that's where we SHOULD be mixing to. Tweakheadz mentions that it's loud, but it doesn't give much more explination beyond that, and anyone that's spent any time in the Studio Central boards knows to not ask "stupid questions" because they'll just ridicule you, often without offering any substantial advice.

And, what dB should I be mixing to, on average? Where should my dialogues sit? Explosions?

In closing, sorry for the harshness of my Dynamic Interference Sound Design Challenge, and for any damages that may have occured because of my inexperience (I did, at least, realize that there was a TON of super loud subbass that was almost inaudible on my monitors, so I high-passed that off...).

Answer 1

Don't worry, Dave, yours wasn't the loudest. ;) 0dB is a mine field, because it all depends on which scale you're referring to. If you're referring to Full Scale (like I did in the post), 0dB is the absolute maximum volume you can produce in a digital system before clipping occurs. Someone suggested that in addition to -10dB full scale, we should have 0db on a VU meter equal -20dB full scale. You see where this is going yet?

There are all kinds of measurement systems, and it all depends on how your meter is calibrated. For a while now, the idea of dialnorm (where the dialogue should sit in the mix) has been ruling audio for film and television measurements. For a long time, the US used a scale known as Linear EQ A, you'll see it abbreviated in a lot of places as LeqA. Using that scale, many broadcasters designated a dialnorm of -27dB LeqA, and you need special equipment to be able to get that measurement. A common dialnorm for film was -31dB LeqA.

That measurement system is going away now, as the new suggested spec is BS.1770 (something that's been gaining ground in Europe for a number of years now), which you'll see referred to as LKFS. A very common measurement for television using an application of this system (i.e. ATSC/A85 or EBU-R128) is -24dB LKFS/LUFS. They've switched over to this system because of arguments regarding the way each scale works.

LeqA really only accounted for dialogue volume over time. The program could get all kinds of loud without going over the -10dbfs (full scale) peak limit spec, as long as all dialogue measured out to -27dB LeqA. The LKFS scale takes into account the rest of the audio program and is an indicator of the program's overall loudness level over time.

This is a lot of technical jargon that doesn't really answer your basic question of where to mix your levels to. It's kind of hard to tell you "x value" dB full scale is where you want to be. A good starting place is to keep all peaks below -10dB full scale. From there, it's a question of how much dynamic range do you want in your piece. You should really start by calibrating your monitors. A quick and dirty way is to play some pink noise through your monitors at -20dbFS and use a dB-SPL meter (you can get these cheap at Radioshack) and adjust the monitors until that meter reads [78dB for TV style mixing/83dB for film style mixing] from your listening position. This means that the max your system will output Full Scale is 98dB-SPL or 103dB-SPL respectively (getting close to the threshold of damage/pain with film mixing, but staying below it). After you've done that play a -20dbFS 1kHz sine wave out through the monitors to give yourself a reference point. Then you can load in something professionally (DVD's can be great for this) done and A/B the program vs. the sine wave, and you'll train your ears for an appropriate level.

This is a better way to start out than to look for "the magic dB" level. If you want to find out how loud an audio file is, I'd suggest loading it into your DAW and using a plug-in that lets you find its RMS value (the gain plug-in in Pro Tools can do this). That will give you yet another reference point. Mixing is all about those reference points and dynamic range. Keep that in mind, and you'll get the hang of what you're looking for in no time.

Edit: I will admit that, when working on television programming, I start out by premixing dialog/narration to a comfortable, but present, level. If I run a semi-QC check on it, it regularly hangs in the -26 to -28 LeqA using Digidesign's Phasescope plug-in (measurement time scale/duration of 3 seconds). I then mix everything else around it.

Answer 2

"0 dB" is unity gain. dB by itself is a relative measurement, not an absolute level. It just refers to a gain unless you give a reference.

Perhaps you're referring to "0 dBFS"? This is the absolute maximum a digital system can measure (clipping), and in this case, your signals should only reach 0 at rare peaks, if ever. The average RMS level should be significantly lower.

Or maybe you mean "0 VU"? This is equivalent to an electrical level of 0 dBu, which is typically about 25 dB below clipping. In this case, the level should generally hover around 0 on the meter.

The right level depends on the crest factor of your signal (Wikipedia says 12–18 dB for a processed mix or 18–20 dB for unprocessed recording), and the dynamic range of the system. If you record too high, you get clipping. If you record too low, you get noise. You need to find the best point between the two.

Answer 3

I have one little question about loudness calibration levels. It seems there are 2 norms:

• -20dB FS of pink noise RMS = 85 dB(C) SPL used for film, 79 for TV (it is 6 dB below film)
• the second is pink noise on 83 dB(C) for film and 78 for TV (only 5 dB below film), even 73 for music (according to Bob Katz K-system).

First is used in Central Europe - especially Czech republic, maybe the rest of Europe is using another calibration levels. Second (I believe) is used in USA.

Where is the truth?

Answer 4

I won't be talking about dB scales but rather about the trick I used for evaluating how loud I should get.

I looked for a DVD with explosions in my library. I chose Predator, the first one, the best one :) I played it and looked for the scene where they attack the rebels camp. I set the DVD Player volume to 100% and my monitors' volume so that it seems as loud as in a theater (well, my speakers are MUCH closer so you know, I mean apparent loudness). While mixing, I would go back and forth between Pro Tools and DVD Player to make sure. The explosions in Predator are really not that loud, but it's dynamic :)

Answer 5

WRT the DI sound design challenges I would suggest that instead of limiting peaks, we agree on a level to monitor at while designing. This way we get to keep the use of all of our dynamic range while still standardizing levels. You know, like they do in film and tv. :)

Even if its not perfect, if everyone blows pink through their rigs as they sound design and set the monitors to 80 or whatever is agreed upon, then mixes to taste we'd all end up in similar spots.

I know the super loud ones in the challenge can not have been designed at a reasonably loud playback level.

Limiting peaks won't solve the apparent loudness issue. Only setting something like -20 dbfs = 80db will. If people come too far off of that, it'll be very obvious.

FWIW I did mine at -20 = 78dbfs, which means it sounds awesome if you're monitoring that way, and it sounds kind of wimpy if you're listening at a level that would make Dave's sound right.

Answer 6

I am from the games industry and not too up to speed on post production standards for film and TV. I understand the need for EQ room calibration when designing and mixing, but I just tested my own work station here and pink noise is -20 dbfs = 65 db SPLc at the listener postion. Turning the system up to 78 SPLc is really fatiguing. Games tend to be about (guesstimate) 10 - 15 db louder than -20 = 78 standard.

So now to my questions - when you guys work on a session do you just level your channels so you are using the full bit depth of the channel, then turn down the master to taiste (giving yourself the option to go louder or quieter), or do you actually mix 'quiet' leaving the master at 0 dbfs, and turn down the channels? (silly question but just curious)

Secondly - and this may be a bit conterversial - if I've undestood this correctly - why are you suggesting using a mix standard of e.g. -20 = 78, by internet video standards this would be quiet? most e.g. vimeo videos are closer to peak 0 dbfs as far as I can tell. IMHO you should really mix to the level that is required for the medium.

btw I am asking more out of curiousity than to cause trouble ;)

Answer 7

Technically, I read - 0,3 db is pretty safe. 100% safe is - 6db. That is the peaks, of course. No peak going above -6db and/or -0.3 db. Don't forget that volume is not really measurable. It is mostly perceived. There is a harmonic balance to be found between high mids and lows. If you have too much lows, you volume will peak and very low perceived level. If you have many highs or high mids, and your mix is peaking constantly at 0,03 db, it might be too loud, because it is perceived as very loud. Don't forget also that the loudness is felt in context. If you are having more volume contrast, the loud passages will of course "feel" louder, but all the rest will feel a lot quieter. So if you are having very little dynamic range, everything will "feel" louder overall. But as often discussed, not necessarily better.

So it's little more complicated than to know your peak levels. I mostly try to work with references to not go overboard in any direction.

Answer 8

It all depends what you're mixing and where is it going to be exhibited. For instance when I am mixing music I keep my peak at -8 dBFS. When it comes to film my loudest mix stem is dialogue which is at -10 dBFS.

You need to understand where your, Mix is going to be exhibited. Keeping that in mind my recording levels are also set that way. During Location recording my levels are more or less at -16 dBFS. sometimes my peaks go up to -14 dbFS and that about it. I do so because it gives me headroom on either side during post production. Even after I know that all my dialogue peaks are going to be at -10 dBFS I still like to keep the headspace so that when I mix i create depth and perception of loudness.

For Music, Due its very nature, you would want to achieve a flat frequency response from your mix. It's easy to achieve that by getting your peaks up to -8 dbFS. If you can achieve it at a lower level then it's great.

Also you should keep in mind the back compatibility of your hardwares and softwares. Assuming that you know that and have your monitor system aligned. Finding our own level range should not be difficult. After all Sound is all perception.

Remember mixing film for DOLBY DIGITAL and DOLBY SR is a Different ball game all together

Answer 9

I am still very confused with 0dB audio. I really kown very well that dB is a unit of gain. When I tested a auido system to confirm if the audio signal was clipped by the audio amplifier, I did as follows: First, I searched a audio file called 0dB/1KHz in form of MP3 on the internet; Second, After Downloaded to my computer, the audio file was stored in SD card; Then, the SD card was inserted into my moblie phone, while the audio player was being turned on to paly the audio file(0dB/1KHz), a oscillograph was being used to measure the amplitude of the signal which was outputing to the loudspeaker.

If the shape of the output signal was clipped very seriously, the gain of the audio amplifier must be decreased.

All above this, What I really want to kown is that. (1)what's the amplitude of the 0dB audio file(in form of MP3) after it went through a 0dB-system? (2)How was the 0dB file(in form of MP3) generated? I thought there must had been a analog natural audio signal first, then people used some audio equipment to caputure this audio signal. After that the signal was converted to digital sigal. Here came the following questions: (3) what's the precision of the A/D converter integrated into the audio equipment? In other words, how many bits did the converter have? Is it a 10-bit A/D converter? or 20-bit? …

I've been looking foward to get the more professional answers about 0dB audio file.

Answer 10

According to the scale on this page, 0 dB (aka 0 dB-SPL) is the sound level in a recording studio:

We said above that the decibel is a ratio. So, when it is used to give the sound level for a single sound rather than a ratio, a reference level must be chosen. For sound pressure level, the reference level (for air) is usually chosen as pref = 20 micropascals (20 μPa), or 0.02 mPa. This is very low: it is 2 ten billionths of an atmosphere. Nevertheless, this is about the limit of sensitivity of the human ear, in its sensitive range of frequency.

0 dB = 20 μPa = 20 μN / m² = 20 μJ / m³

Anyway, you'll probably want to avoid clipping, so should stay below 0 dBFS, the loudest level:

dBFS, or decibels relative to Full Scale, is used to measure digital audio signal levels. dBFS is another dimensionless quantity, because it is just a number and cannot be converted to another unit. In a digital audio system, 0dBFS refers to the maximum signal level possible, also known as the clipping point. Therefore, dBFS values are always less than or equal to zero. -10dBFS corresponds to a signal that is 10dB lower than the clipping point of the system.

To make your recording sound louder/"better" overall, you could scale down the peaks instead of clipping them, but that still loses fidelity. See The Loudness War.

Answer 11

0dB is the highest any peaks can go, before it clips. I recommend mixing your music to about --10dB average and 0dB peak. Anything quieter will annoy anyone who's using a laptop - they'll feel like they can "just barely hear it".