Reference sound pressure level flowchart

[flipflop]

Why use a reference SPL?
"The reference listening level is defined as a preferred listening level, produced with a given measuring signal at the reference listening point. It characterizes the acoustic gain of the reproduction channel in order to ensure the same sound pressure level in different listening rooms for the same excerpt."
"Reference level is a calibrated volume setting used for both movie production (in dubbing stages and post production houses) and reproduction (in screening rooms and theaters). The human hearing system is non-linear, especially in the bass, so having a consistent playback level – a reference – is critical if the mix is to translate from one production house to another and audiences are to hear the director’s intent in terms of the balance in the soundtrack between dialog, effects and ambiance."

Calibrating loudspeakers
1. Copy the first test tone, 1 kHz sine wave at -20 dBFS, to a digital audio workstation, and then copy it into each of the channels to be employed.
2. Play the file, setting the output level controls of the workstation to unity gain, the console input to unity gain, and the console master level control to unity gain. You may keep the monitor level control low for the time being.
3. Set the level of each channel in turn for the sine-wave test level for the meter in use, such as -20 dBFS for digital meters, or 0 VU for VU meters.
4. Copy the second test tone, band-limited pink noise at -20 dBFS, to each channel of the digital audio workstation.
5. Play one channel at a time if calibrating for music or television. Play all channels except the subwoofer (which is calibrated separately) if calibrating for movies.
6. Put a sound level meter at the position of the center of the head of the normal position for the operator, pointing it at the loudspeaker channel in use. Keep your body perpendicular to the meter, off to one side, as the level can be affected by a strong reflection off your body.
7. Set the individual channel level controls, such as power amplifier gain controls, or gain controls on powered loudspeakers for the Sound Pressure Level measured C-weighted and slow.

Or put more simply, using playback for movies as an example:
"Reference level for all channels except low frequency effects is calibrated by adjusting the audio chain such that a pink noise signal recorded at -20dB relative to full scale (0dB) creates 85dB sound pressure level as measured with a C weighted SPL meter at the seating locations."

Flowchart

*This is 80 dB @ 1 kHz if the headphones follow the Harman target curve.

Sources
[1] Dolby 5.1-Channel Music Production Guidelines
http://www.voesd.at/files/Multichannel_Music_Mixing.pdf

[2] Methods for the subjective assessment of small impairments in audio systems
Recommendation ITU-R BS.1116-3 (02/2015)
https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.1116-3-201502-I!!PDF-E.pdf

[3] A Statistical Model that Predicts Listeners’ Preference Ratings of Around-Ear and On-Ear Headphones
http://www.aes.org/e-lib/browse.cfm?elib=19436

[4] A Statistical Model That Predicts Listeners’ Preference Ratings of In-Ear Headphones: Part 1 – Listening Test Results and Acoustic Measurements
http://www.aes.org/e-lib/online/browse.cfm?elib=19237

[5] RP 200:2012 - SMPTE Recommended Practice - Relative and Absolute Sound Pressure Levels for Motion-Picture Multichannel Sound Systems — Applicable for Analog Photographic Film Audio, Digital Photographic Film Audio and D-Cinema
https://ieeexplore.ieee.org/document/7289847

[6] ATSC Recommended Practice: Techniques for Establishing and Maintaining Audio Loudness for Digital Television (A/85:2013)
https://www.atsc.org/wp-content/upl...tablishing-and-maintaining-audio-loudness.pdf

 

[pozz]

Nice.

In certain examples, why A or C weighting rather than Z (flat)? To address room effects?

 

[MRC01]

Nice chart but seems on the loud side all around. If -20 dB is at 85 dB SPL then it can get up to 105 dB SPL. The NIOSH max safe exposure at that level is about 4 minutes. At these levels one could easily exceed that. I find movie theaters much too loud and wear custom mold earplugs with Etymotic ER-15 dB filters. Also it's advisable to set headphones not louder than speakers but quieter; the risk of playing too loud is higher with headphones since they don't have some the loudness cues that speakers have. Finally, if you play things loud, it triggers your ear's biological dynamic compression to compensate, which can mask subtle fine detail.
Personally, I would reduce all of the boxes by 3 to 6 dB.

 

[Blumlein 88]

Need to note 85 db SPL is two speakers. Or alternatively one can use one speaker at 83 db SPL with pink noise. Also are the volumes for listening room cubic meters or feet?

For those complaining it is too loud, Dolby did a bunch of research to arrive at this. Bob Katz in explaining his K-system for mastering bases it upon this Dolby work. He describes in a demo at I think the AES how they played some tracks based upon this level in an auditorium. When done asked how many think this was too loud, about 5 hands shot up. How many think it is too quiet, no hands. How many thing it was just right, and about 400 hands went up.

Now I've found setting up stereo playback rigs or home theaters that this system is nigh on perfect for movies. Usually is for music if it isn't mastered to death. Network TV sometimes violates these guidelines.

In any case, for most of us setting up this for a 76 db reference level for one speaker, on movies and well done music you'll want to adjust it at most 3 db either way for like 99% of what you'll run across. The NIOSH exposure limit of 105 db SPL for 4 minutes won't apply (nor will your ears be harmed) because it is for continuous noise exposure levels. In movies and well mastered music, you'll hit and recede from these levels fleetingly. Surprisingly even Michael Bay movies while they'll seem too loud too often, they follow these guidelines and won't leave your ears blasted. You'll just think the dude is over the top on explosions.

My comments are in regard to speakers, not headphones btw.

 

[mitchco]

@flipflop Very nice! This is indeed "reference level" per the standards and what I used to mix sound for the final tonal balance. I am definitely not the only one that does this as it is (and has been) taught in audio engineering schools for decades. You might consider this as a reference as well. Bob Katz is a well known mastering engineer and has been on this for years: https://www.digido.com/portfolio-item/level-practices-part-2/ For most producing music and movies, Bob's K-System is the unofficial defacto standard.

Another standard to add to your list:
EBU Tech 3276, “Listening conditions for the assessment of sound programme material: monophonic and two–channel stereophonic."
It is pretty much the same as the ITU, but another to add. You can also hunt down the Dolby 5.1. Mix guidelines for music as well if you can find it. I attached an old copy here.

I am sure you know this but here is a high level explanation with some audio samples to show our ears frequency response is not only non-linear, but also changes with SPL: The non-linearities of the Human Ear

Part of the reason that some music has more bass response than others is directly related to what level it was mixed/mastered at. Too low of a monitoring level (e.g. 65 dB SPL) and in the mix, you inherently boost the bass/treble to compensate for our ears falling response at both frequency extremes. Mixing at too loud of a level (e.g. 95 dB SPL i.e. “concert level”) and you instinctively turn down the bass and treble in the mix, as at that SPL our ears sensitivity to bass and treble has increased considerably in relation to other frequencies.

The irony is some rock records I listened to in the 80’s had no bass and treble as they were mixed/mastered at concert level and sounded wimpy at normal (i.e. reference) listening level. I am guilty of this myself until I got educated.

I also love using loudness controls for when listening below "reference level." Here is a good example of one in JRiver including the calibration procedure: https://yabb.jriver.com/interact/index.php?topic=76608.0 You might consider adding that to your fow chart of listening below reference level.

The reason why you don’t see loudness controls too much as they actually need to be calibrated with Pink Noise and a sound pressure level (SPL) meter, as linked in my previous post, for it to mean anything. If you do calibrate for playback, then you also need to enable volume levelling (i.e. EBU R128) so that each song you are playing back is at the calibrated level, or you will be manually adjusting the volume with the SPL meter for each song.

Cheers!

 

[mitchco]

lol @Blumlein 88 Dennis, we just posted about Bob Katz at the same time

 

[Blumlein 88]

lol @Blumlein 88 Dennis, we just posted about Bob Katz at the same time

Yes, I hit return and then look and see an Alert notice. Hopefully great minds think alike!

 

[MRC01]

... For those complaining it is too loud, Dolby did a bunch of research to arrive at this. ... When done asked how many think this was too loud, about 5 hands shot up. How many think it is too quiet, no hands. How many thing it was just right, and about 400 hands went up. ... The NIOSH exposure limit of 105 db SPL for 4 minutes won't apply (nor will your ears be harmed) because it is for continuous noise exposure levels. ...

I guess I'm one of those 5. I have always had a low tolerance for loud sounds, since I was a kid. If I watch a 2 hour movie in a theater without hearing protection, my ears are ringing when I walk out. That's too loud - at least for me. I wear those same ER-15 earplugs when playing piccolo and playing in large groups.

Regarding exposure limits to avoid hearing damage: it depends on how long is the interval between the 100+ dB peaks, and what is the SPL during that interval. I've seen action movies with extended scenes that push the limits. I have enough friends with noise related hearing loss to know that all this exposure adds up over time.

 

[MRC01]

... I am sure you know this but here is a high level explanation with some audio samples to show our ears frequency response is not only non-linear, but also changes with SPL: The non-linearities of the Human Ear
...

I believe this is why so many audiophiles listen at levels that are louder than reality - especially for small ensemble acoustic music. The louder level is like boosting the bass & treble. Which is ironic, since many of these same audiophiles eschew the use of tone controls, not realizing their volume knob is a kind of tone control.

 

[flipflop]

In certain examples, why A or C weighting rather than Z (flat)?

To be honest, I don't understand why weighting is used at all for this purpose.

Need to note 85 db SPL is two speakers.

True enough, the citation does mention "all channels except low frequency effects". I've updated step 5 to reflect this fact.

Or alternatively one can use one speaker at 83 db SPL with pink noise.

Are you talking about a mono set-up here? Because 83 dB + 83 dB = 89 dB.

Also are the volumes for listening room cubic meters or feet?

Cubic feet. Read the chart closely

@flipflop Very nice! This is indeed "reference level" per the standards and what I used to mix sound for the final tonal balance. I am definitely not the only one that does this as it is (and has been) taught in audio engineering schools for decades. You might consider this as a reference as well. Bob Katz is a well known mastering engineer and has been on this for years: https://www.digido.com/portfolio-item/level-practices-part-2/ For most producing music and movies, Bob's K-System is the unofficial defacto standard.

Thank you. I'm familiar with Bob Katz, the K-System, and also the article. It was discussed in the thread 'SPL for critical listening', which is my inspiration for this thread. Lots of different sources were being cited and it became clear to me that because the reference level depends on playback content, listening equipment, and even room size, the most sensible way of answering the question of what SPL to use for hi-fi listening would be a flowchart, as there's more than just one single answer.

Another standard to add to your list:
EBU Tech 3276, “Listening conditions for the assessment of sound programme material: monophonic and two–channel stereophonic."
It is pretty much the same as the ITU, but another to add.

This standard uses the same formula (Lref = 85 log(n) dB(A)) as ITU-R BS.1116-1 and BS.1116-2.
The newest iteration (ITU-R BS.1116-3) simply states that Lref = 78 dBA, which led me to believe that the formula incorporating the number of channels is outdated.
https://www.itu.int/rec/R-REC-BS.1116/en

You can also hunt down the Dolby 5.1. Mix guidelines for music as well if you can find it. I attached an old copy here.

I did look around for a Dolby citation for the 85 dB movie reference level because I felt my sources were a bit weak. Didn't know Dolby also had guidelines for music. Thanks a lot, will have a look at it tomorrow.

I am sure you know this but here is a high level explanation with some audio samples to show our ears frequency response is not only non-linear, but also changes with SPL: The non-linearities of the Human Ear

Yes, the problem reference levels tackle is psychoacoustic at its core, hence my decision to create this thread in the 'Psychoacoustics' sub-forum. Appreciate the link.

Part of the reason that some music has more bass response than others is directly related to what level it was mixed/mastered at. Too low of a monitoring level (e.g. 65 dB SPL) and in the mix, you inherently boost the bass/treble to compensate for our ears falling response at both frequency extremes. Mixing at too loud of a level (e.g. 95 dB SPL i.e. “concert level”) and you instinctively turn down the bass and treble in the mix, as at that SPL our ears sensitivity to bass and treble has increased considerably in relation to other frequencies.
The irony is some rock records I listened to in the 80’s had no bass and treble as they were mixed/mastered at concert level and sounded wimpy at normal (i.e. reference) listening level. I am guilty of this myself until I got educated.

Very reminiscent of the circle of confusion.

I also love using loudness controls for when listening below "reference level." Here is a good example of one in JRiver including the calibration procedure: https://yabb.jriver.com/interact/index.php?topic=76608.0 You might consider adding that to your fow chart of listening below reference level.

Loudness/tone control is definitely a good alternative to correct playback levels. The thread 'Device for Equal Loudness/Fletcher Munson Curve? Do Any Speakers Adapt to This?' discusses a lot of options and also served as a reminder for me to make this one. I think I would like to keep the topics separate for now.

 

[MRC01]

... Are you talking about a mono set-up here? Because 83 dB + 83 dB = 89 dB. ...

Twice the power is 3 dB, so 2 speakers each individually (with the other turned off) generating 83 dB, should be 86 dB with both playing.

 

[flipflop]

I updated the flowchart with the reference SPL for 5.1-channel music and a new look. I hope it's easier to read now.

 

[jlo]

View attachment 47920
In this diagram, the value for (>Movies>Speakers) is wrong : it should be 89-90dBC for sub. Because the standard asks for an RTA measurement 10dB higher for LFE compared to a front loudspeaker, that is about 89 to 90dBC depending on subwoofer frequency response.
Another detail : in pro applications, SPL calibration should be done with full band pink noise (ie SMPTE ST2095-1 wich has a measured rms level near -22dBFSD)

 

[jlo]

Twice the power is 3 dB, so 2 speakers each individually (with the other turned off) generating 83 dB, should be 86 dB with both playing.

In a room when you measure SPL of two loudspeakers together feeded with same wideband noise, the SPL gets generally near +4dB, because the addition is non-coherent (uncorrelated) in higher frequencies (+3dB) but becomes more coherent in lower frequencies (+6dB)

 

[andymok]

There's a great book out there

Audio Metering: Measurements, Standards and Practice (Audio Engineering Society Presents)

2nd Edition
https://www.amazon.com/Audio-Meteri...keywords=audio+metering&qid=1580550930&sr=8-1

3rd Edition
https://www.amazon.com/Audio-Meteri...keywords=audio+metering&qid=1580550930&sr=8-2

 

[flipflop]

In this diagram, the value for (>Movies>Speakers) is wrong : it should be 89-90dBC for sub. Because the standard asks for an RTA measurement 10dB higher for LFE compared to a front loudspeaker, that is about 89 to 90dBC depending on subwoofer frequency response.

That is the case with 5.1-channel music. It's clearly stated, in the article cited for Movies+Speakers, that the sub must measure 95 dBC with a SPL meter. If you have a more authorative source that contradicts this, please provide it.

Another detail : in pro applications, SPL calibration should be done with full band pink noise (ie SMPTE ST2095-1 wich has a measured rms level near -22dBFSD)

SMPTE ST2095-1 isn't publicly available, so I can't comment on it. Just about all the sources I've provided mention -20 dBFS.

There's a great book out there

Audio Metering: Measurements, Standards and Practice (Audio Engineering Society Presents)

2nd Edition
https://www.amazon.com/Audio-Meteri...keywords=audio+metering&qid=1580550930&sr=8-1

3rd Edition
https://www.amazon.com/Audio-Meteri...keywords=audio+metering&qid=1580550930&sr=8-2

Thank you. I'll check out the 3rd edition after its release in May.

 

[jlo]

That is the case with 5.1-channel music. It's clearly stated, in the article cited for Movies+Speakers, that the sub must measure 95 dBC with a SPL meter. If you have a more authorative source that contradicts this, please provide it.

In all SMPTE standards, it is written that the RTA measured value of LFE, and not the SPL value, should be 10dB higher : here is copy of the latest version (2012)

SMPTE ST2095-1 isn't publicly available, so I can't comment on it. Just about all the sources I've provided mention -20 dBFS.

From 25CSS-DP-ST2095-1-Pink Noise-20150811, Calibration Reference Wideband Digital Pink Noise Signal :
0 dB FS - The RMS level of a 997 Hz sine wave whose positive peak value reaches Full Scale Digital. (Note: This may be depicted as “0 dBFS” in some literature. The meaning is the same.) So rms value of 0dBFS is -3dBFSD.
FSD - Full Scale Digital. The maximum numeric value capable of being stored in the bit depth of a digital signal path. Example: A full scale sine wave has positive peaks with a value of 7FFFFF (hex) and negative peaks with a value of 800000 (hex) in a 24 bit digital audio system.

This explains why the rms value of the pink noise is -19dBFS -3dB = -22dBFSD (Full Scale Digital)

This surprising value of -21.5dBFSD unfiltered (or -22dBFSD filtered 22-22k) comes from ancient times : it has been chosen to be very near to the old Dolby analog and digital pink noises (unfiltered levels at -21.5FSD).
And those levels comes from an even older time where the alignement level was set 20dB below the maximum level. But what is 20dB below the max level of a 1khz sinus signal ? Is it 20dB below the max peak (0dB) or below the max rms (-3dB) ? Strangely, the historical value was just in between !

 

[flipflop]

In all SMPTE standards, it is written that the RTA measured value of LFE, and not the SPL value, should be 10dB higher : here is copy of the latest version (2012)

Thank you. I've updated the flowchart now.
@mitchco, please speak up if you have any objections.

From 25CSS-DP-ST2095-1-Pink Noise-20150811, Calibration Reference Wideband Digital Pink Noise Signal :
0 dB FS - The RMS level of a 997 Hz sine wave whose positive peak value reaches Full Scale Digital. (Note: This may be depicted as “0 dBFS” in some literature. The meaning is the same.) So rms value of 0dBFS is -3dBFSD.
FSD - Full Scale Digital. The maximum numeric value capable of being stored in the bit depth of a digital signal path. Example: A full scale sine wave has positive peaks with a value of 7FFFFF (hex) and negative peaks with a value of 800000 (hex) in a 24 bit digital audio system.

This explains why the rms value of the pink noise is -19dBFS -3dB = -22dBFSD (Full Scale Digital)

This surprising value of -21.5dBFSD unfiltered (or -22dBFSD filtered 22-22k) comes from ancient times : it has been chosen to be very near to the old Dolby analog and digital pink noises (unfiltered levels at -21.5FSD).
And those levels comes from an even older time where the alignement level was set 20dB below the maximum level. But what is 20dB below the max level of a 1khz sinus signal ? Is it 20dB below the max peak (0dB) or below the max rms (-3dB) ? Strangely, the historical value was just in between !

I appreciate the technical explanation and the history lesson, but this seems way too convoluted to be used in a practical manner.

 

[CDMC]

Bringing this back up. Great idea and flow chart. A couple of what I believe are errors:

1) For your pink noise, you should be using 500-2000hz pink noise to eliminate errors when calibrating due to variances in speaker responses at a specific frequency.
2) Your flow chart does not account for room size when listening to stereo and will result in too loud of levels for most people. See the chart at the bottom of the page. https://www.soundonsound.com/techni...studio-reference-monitoring-levels?page=3#top
3) You should also calibrate for -14db (LUFS). This is because that is what most streaming services use. If you don't, you will end up with levels +6db for streaming listening (note that Roon allows you to select your volume leveling).

 

[Senior NEET Engineer]

The TV SPL looks off. 20,000+ cu ft... what kind of use case is that for TV... that is a movie use case. Also note that TV shows use very high dynamic range compression. I turn the volume down 10-20 dB compared to movies.