Fletcher Munson Curve verses a flat frequency response.

[Ashley Salmond]

Dear fellow Audiophiles. I happen to to be watching YouTube where a black American had the good fortune of comparing the latest edition of the Arendal speakers to the same size Perliston and which one had the flattest frequency response in his listening room in which the Perlisten was clearly winner. What did surprise me was the Arendal latest edition had a similar response to my Thx tower S 1726. I then came across a well known Audio guy talking about the Fletcher Munson curve, and how similar the Arendal speakers match that curve in room . I have seen on Audioholics where a regular guest ( with a beard) was getting rid of his Revel Performers for a set of Focals and complained about the Revels lack of bass. No these Revels speakers have a very flat frequency response. I guess my question is - what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening .

 

[boxerfan88]

If we look at the FM curve, they crunch together in the bass area.

That kinda lines up with my own experience.

At moderate volume (eg. ~75dBSPL) I prefer more bass (approx 3-5dB boost).

At high volume (eg. ~90dBSPL) I prefer flatter bass (ie. less boost to no bass boost)

 

[Sal1950]

No these Revels speakers have a very flat frequency response. I guess my question is - what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening .

IMHO it's always best to have a speaker capable of a flat frequency response (at least close to it).
You can always modify any way that pleases you with tone controls, DRC, whatever.

BTW the F-M curve is mostly considered obsolete
Perceived discrepancies between early and more recent determinations led the International Organization for Standardization (ISO) to revise the standard curves in ISO 226. They did this in response to recommendations in a study coordinated by the Research Institute of Electrical Communication, Tohoku University, Japan. The study produced new curves by combining the results of several studies—by researchers in Japan, Germany, Denmark, UK, and the US. (Japan was the greatest contributor with about 40% of the data.)

 

[kemmler3D]

what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening .

So, I may be butchering this a little, but Fletcher Munson curves describe how loud a tone of a given frequency sounds to an average person at a given SPL. As it happens, the same SPL won't sound equally loud at every frequency, you have to bump it up especially in the low bass and high treble - our ears are less sensitive there.

So the FM curves describe how sensitive the ear is. It seems intuitive that speakers should account for this, right?

Actually, no! Because the sensitivity of the ear is already at work when the recording is made. Whatever variations we get from the FM curves are already baked into the recording by the people who made it. They've accounted for the fletcher munson curves in the recording itself. So altering the frequency response again, to account for the FM curves a second time, would be redundant and actually just adding distortion.

This is why a flat frequency response is desirable, ideally it simply reproduces what the artists intended you to hear.

Where FM curves do come into play during playback is when you are listening quietly or very loud. As you'll note the FM curves are shaped differently depending on the SPL. Our ears' frequency response changes depending on volume. This is called the loudness effect. Annoying!

As such, changing the frequency response to account for loudness effect if you are listening at quiet or loud volumes can be useful. This is what the "loudness" control on some preamps and receivers does.

You can also make an argument that yes, the speakers should have a non-flat frequency response for similar reasons, or perhaps you just prefer the sound that way.

But to tie it back to your original question, a speaker with a flat response is not failing to account for anything.

As for lack of bass, bass is pretty variable depending on the room and it is more than acceptable to EQ it for room correction or even just personal preference. I think this is neither here nor there in deciding whether you want a speaker with a flat response.

 

[respice finem]

Equal loudness curves (if FM or other) are actually multiple curves dependent on loudness, not one static curve. IDK how a loudspeaker could adhere to them "by itself", and why should it even attempt to. We are living in the "DSP era" and not in the 70s or 80s, so DSP is the way to achieve proper loudness equalization, not buying "coloured" loudspeakers.

Equal-loudness contour - Wikipedia

en.wikipedia.org

 

[Sal1950]

I guess my question is - what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening .

After a second reading of your post the final question confuses me?
No speaker, specially studio monitors have a response designed to match the F-M curves, The F-M curve was created after measuring human hearing and coming up with a series of response modifications to make bass sound equally loud at various playback levels. You can see the same happens on the top end though not as drastically as the bottom. (unless your old like me)

 

[smygolf]

After a second reading of your post the final question confuses me?
No speaker, specially studio monitors have a response designed to match the F-M curves, The F-M curve was created after measuring human hearing and coming up with a series of response modifications to make bass sound equally loud at various playback levels. You can see the same happens on the top end though not as drastically as the bottom. (unless your old like me)

I guess it comes down to production and mastering.
Speakers used could be flat but the material has been Eq:d

 

[Robin L]

Yamaha used to have (maybe still has) these interesting variable loudness controls that shifted the Fletcher-Munson curve depending on the volume level—decrease the volume and the bass increases proportionally. I recall using one of the Yamaha receivers with this control, much better than a typical "loudness" button. I like bass and I probably like it too much, so I like more of it as the level goes down.

 

[Sal1950]

Yamaha used to have (maybe still has) these interesting variable loudness controls that shifted the Fletcher-Munson curve depending on the volume level—decrease the volume and the bass increases proportionally. I recall using one of the Yamaha receivers with this control, much better than a typical "loudness" button. I like bass and I probably like it too much, so I like more of it as the level goes down.

Although the use of loudness controls or buttons goes in and out of popularity with audiophiles, being a bass lover myself I still believe in them.
There's no denying the way our hearing sensitivity changes with SPL, so how can anyone not desire a properly designed compensation plan?
Though far from perfect I still use Audyssey's DEQ, just wish they'd listen to all the input they get from they're customers about improvements.
YMMV

 

[ZolaIII]

You do want speakers which have (predicted) natural in room response as close to our hearing capabilities without any compensation. That's 85/88 dB SPL mono/stereo white noise calibration point (to listening area). Only then when system is properly calibrated you can apply automatic or manual equal loudness compensation to desired listening program SPL (with a help of sub or sub's of course). Why? So that you can have full octave perceived spectrum on sane SPL levels or for quite listening sessions. Most folks won't be able to enjoy it any other way thanks to living environment so ELC comes as saving grace there.

 

[sigbergaudio]

It's important to separate frequency response as measured anechoically (what manufacturers often present in their data, if they share such information), and frequency response as measured in a listening room at the listening position.

Anechoic response will typically be and should typically be pretty flat (no tilt) and even (not a lot of deviations).

In-room response will typically have and should typically have a slope (more energy in the bass).

Something like this:

And/Or this:

 

[Sir Sanders Zingmore]

a black American

sorry, what?
How is the person’s skin colour even vaguely relevant here?

 

[Chris A]

I guess my question is - what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening

I think you are missing one critical piece of the "system": the music (or speech) itself--which already has equal loudness curves baked into them.

One of the tasks that the music mastering guys perform is taking the mixdown tracks and turning them into listenable music at some reference loudness playback level. They generally use flat transfer function response (amplitude and phase) studio monitors to do this. When they don't use loudspeakers having at least flat amplitude response, there are problems with playing the music back on our "consumer grade" loudspeakers. Generally this is perceived as "bass heavy" or "bass shy", but too much or too little high frequencies are equally as likely.

Here is a JAES article on the subject of "Programme Material Analysis" by Chapman (B&O) that shows the mean amplitude responses of the mastered music tracks by genre (enclosed below).

However, when you choose to play your music back at a loudness that doesn't match the conditions of the mastering loudness levels (generally, this is ~83 dB at the mastering guy's ears), then there are differences that show up based on high and/or low frequency levels that are generally EQed to a non-flat amplitude level to approximate the equal loudness curves for the level that you are listening to the music (with some special EQ applied around the room's Schroeder frequency where a half-wavelength of sound generally just fits the room dimensions).

What's critical, however, is that at least the amplitude response of your playback speakers exhibits flat response (and having flat phase response is also a factor in how the music is perceived).

Chris

 

[Hayabusa]

So, I may be butchering this a little, but Fletcher Munson curves describe how loud a tone of a given frequency sounds to an average person at a given SPL. As it happens, the same SPL won't sound equally loud at every frequency, you have to bump it up especially in the low bass and high treble - our ears are less sensitive there.

So the FM curves describe how sensitive the ear is. It seems intuitive that speakers should account for this, right?

Actually, no! Because the sensitivity of the ear is already at work when the recording is made. Whatever variations we get from the FM curves are already baked into the recording by the people who made it. They've accounted for the fletcher munson curves in the recording itself. So altering the frequency response again, to account for the FM curves a second time, would be redundant and actually just adding distortion.

This is why a flat frequency response is desirable, ideally it simply reproduces what the artists intended you to hear.

Where FM curves do come into play during playback is when you are listening quietly or very loud. As you'll note the FM curves are shaped differently depending on the SPL. Our ears' frequency response changes depending on volume. This is called the loudness effect. Annoying!

As such, changing the frequency response to account for loudness effect if you are listening at quiet or loud volumes can be useful. This is what the "loudness" control on some preamps and receivers does.

You can also make an argument that yes, the speakers should have a non-flat frequency response for similar reasons, or perhaps you just prefer the sound that way.

But to tie it back to your original question, a speaker with a flat response is not failing to account for anything.

As for lack of bass, bass is pretty variable depending on the room and it is more than acceptable to EQ it for room correction or even just personal preference. I think this is neither here nor there in deciding whether you want a speaker with a flat response.

perfect!

 

[DanielT]

If you ONLY listen at low volume, you may find it sensible to adjust the Fletcher Munson curve to your listening level. If the speakers themselves have such an FR or, easier then to find specific types of speakers, you fix it via EQ, it doesn't matter. Test and see if it's something you like.

If you play at really low volume, you will notice how little bass and treble there is in what you hear. It is clearly audible. Everyone has noticed this FM curve effect, although perhaps not something you think about.

If you listen at different volumes, you should of course not, even if there are speakers with really high bass and treble boost in the FR, choose such a speaker. Then you have to compensate in the other direction, so to speak, when you play at normal/high volume.
It is best to choose a neutral speaker and then EQ if necessary and desired.
That said, and now I may be contradicting myself, but speakers that have a little extra boost in the bass area can be nice to listen to. Taste and opinion and all that.....Sorry, that was a bit of a waffle on my part there.

Loudness/contour compensation functionality on amplifiers was common in the past:


Why did that feature disappear on modern amplifiers? Don't people listen at lower volumes anymore?
I find it convenient to press a loudness/contour button when I'm listening at low volumes anyway.

A side note. In the bass frequency range FR can look and sound crazy due to the room's impact, so EQ may make sense to perform in any case.

 

[respice finem]

If you ONLY listen at low volume, you may find it sensible to adjust the Fletcher Munson curve to your listening level. If the speakers themselves have such an FR or, easier to find specific types of speakers, you fix it via EQ, it doesn't matter. Test and see if it's something you like.

If you play at really low volume, you will notice how little bass and treble there is in what you hear. It is clearly audible. Everyone has noticed this FM curve effect, although perhaps not something you think about.

If you listen at different volumes, you should of course not, even if there are speakers with really high bass and treble boost in the FR, choose such a speaker. Then you have to compensate in the other direction, so to speak, when you play at normal/high volume.
It is best to choose a neutral speaker and then EQ if necessary and desired.
That said, and now I may be contradicting myself, but speakers that have a little extra boost in the bass area can be nice to listen to. Taste and opinion and all that.....Sorry, that was a bit of a waffle on my part there.

Loudness/contour compensation functionality on amplifiers was common in the past:
View attachment 473265

Why did that feature disappear on modern amplifiers? Don't people listen at lower volumes anymore?
I find it convenient to press a loudness/countur button when I'm listening at low volumes anyway.

Even with vintage (or "oldschool") amps, it's IMHO better to have a not perfect loudness option, than to have none.
With modern DSP enabled ones, it should be a "no brainer", but is still often omitted, despite the chips in the devices having such capability.

 

[DanielT]

Even with vintage (or "oldschool") amps, it's IMHO better to have a not perfect loudness option, than to have none.
With modern DSP enabled ones, it should be a "no brainer", but is still often omitted, despite the chips in the devices having such capability.

I suspect that for example, via WiiM (which seems to be the trend that many people are increasingly using) you can add different EQ settings. It should be possible to run sweeps at low volume and then activate the Fletcher Munson curve at low volume compensated EQ setting when needed.
However, there is no FM compensation in sync with volume changes in that case, which is actually what would be optimal to have.

A little OT:
By the way WiiM's autocorrection functionality is now also available in WiiM Mini. I don't know how good or accurate it is. Maybe decent? But using your cell phone, as my friend does for his WiiM sweeps, with an uncalibrated cell phone microphone. It works but it can hardly be optimal?

 

[Beave]

Dear fellow Audiophiles. I happen to to be watching YouTube where a black American had the good fortune of comparing the latest edition of the Arendal speakers to the same size Perliston and which one had the flattest frequency response in his listening room in which the Perlisten was clearly winner. What did surprise me was the Arendal latest edition had a similar response to my Thx tower S 1726. I then came across a well known Audio guy talking about the Fletcher Munson curve, and how similar the Arendal speakers match that curve in room . I have seen on Audioholics where a regular guest ( with a beard) was getting rid of his Revel Performers for a set of Focals and complained about the Revels lack of bass. No these Revels speakers have a very flat frequency response. I guess my question is - what is the point of having a set of speakers that have a flat frequency response and no base simply because it doesn’t match the Fletcher Munson curve. I understand it’s good for studio recording but fails to please in room . listening .

You sure about that? You don't want the in-room response to be flat. You want it to slope downward. You want the *anechoic, on-axis response* to be flat.

 

[diablo]

Loudness/contour compensation functionality on amplifiers was common in the past:
View attachment 473265

Why did that feature disappear on modern amplifiers? Don't people listen at lower volumes anymore?
I find it convenient to press a loudness/contour button when I'm listening at low volumes anyway.

About 20 years ago I was looking for a moderately priced integrated amplifier for the lounge. It seemed that nearly all the ones on offer didn't even feature bass and treble controls, let alone loudness control.
As I mainly used to listen to classical music at low volumes this wasn't ideal.
I don't remember any of the hifi outfits offering Yamaha amps at the time. Though eventually found a Harmon Kardon with treble/bass - plus a remote control!

 

[OCA]

I had written the below function for volume adjustment of speakers on a given target curve in A1 Evo. It's for use with REW API but can easily be adapted. It's an accurate interpretation of the ISO 226-2023 paper and could be useful to those interested:

/**
 * ISO 226:2023 Equal-Loudness-Level Contour Model.
 * This object contains the data and methods to calculate the Sound Pressure Level (SPL)
 * for a given loudness level (phon) and frequency, based on the international standard.
 */
const ISO226_MODEL = {
    // Data from ISO 226:2023, Table 1
    frequencies: [20, 25, 31.5, 40, 50, 63, 80, 100, 125, 160, 200, 250, 315, 400, 500, 630, 800, 1000, 1250, 1600, 2000, 2500, 3150, 4000, 5000, 6300, 8000, 10000, 12500],
    alpha_f: [0.635, 0.602, 0.569, 0.537, 0.509, 0.482, 0.456, 0.433, 0.412, 0.391, 0.373, 0.357, 0.343, 0.330, 0.320, 0.311, 0.303, 0.300, 0.295, 0.292, 0.290, 0.290, 0.289, 0.289, 0.289, 0.293, 0.303, 0.323, 0.354],
    L_U: [-31.5, -27.2, -23.1, -19.3, -16.1, -13.1, -10.4, -8.2, -6.3, -4.6, -3.2, -2.1, -1.2, -0.5, 0.0, 0.4, 0.5, 0.0, -2.7, -4.2, -1.2, 1.4, 2.3, 1.0, -2.3, -7.2, -11.2, -10.9, -3.5],
    T_f: [78.1, 68.7, 59.5, 51.1, 44.0, 37.5, 31.5, 26.5, 22.1, 17.9, 14.4, 11.4, 8.6, 6.2, 4.4, 3.0, 2.2, 2.4, 3.5, 1.7, -1.3, -4.2, -6.0, -5.4, -1.5, 6.0, 12.6, 13.9, 12.3],
    
    // Helper to find interpolation indices and factor for a given frequency
    _getInterp(freq) {
        if (freq <= this.frequencies[0]) return { i0: 0, i1: 0, factor: 1 };
        if (freq >= this.frequencies[this.frequencies.length - 1]) return { i0: this.frequencies.length - 1, i1: this.frequencies.length - 1, factor: 1 };
        const i1 = this.frequencies.findIndex(f => f >= freq);
        const i0 = i1 - 1;
        // Logarithmic interpolation for frequency
        const f0 = this.frequencies[i0];
        const f1 = this.frequencies[i1];
        const factor = (Math.log(freq) - Math.log(f0)) / (Math.log(f1) - Math.log(f0));
        return { i0, i1, factor };
    },

    // Helper to get an interpolated parameter value for a given frequency
    _getParam(freq, paramArray) {
        const { i0, i1, factor } = this._getInterp(freq);
        if (i0 === i1) return paramArray[i0];
        // Linear interpolation for parameters
        return paramArray[i0] * (1 - factor) + paramArray[i1] * factor;
    },

    /**
     * Calculates the required Sound Pressure Level (SPL) for a given loudness (in phon)
     * at a specific frequency.
     */
    getSpl(phon, freq) {
        const af = this._getParam(freq, this.alpha_f);
        const Lu = this._getParam(freq, this.L_U);
        const Tf = this._getParam(freq, this.T_f);
        const B = Math.pow(4e-10, 0.3 - af);
        const C = Math.pow(10, 0.072);
        const term1 = Math.pow(10, 3 * phon / 100) - C;
        if (term1 < 0) return Tf; // Below threshold, SPL is the threshold of hearing
        const term2 = B * term1 + Math.pow(10, af * (Tf + Lu) / 10);
        const spl = (10 / af) * Math.log10(term2) - Lu;
        return spl;
    }
};

/**
 * Finds the starting index in a frequency array for a given frequency value.
 * @param {number} freq The frequency to find.
 * @param {Float32Array|number[]} frequencies The sorted array of frequencies.
 * @returns {number} The index of the first frequency greater than or equal to the target.
 */
function freqToIndex(freq, frequencies) {
    const index = frequencies.findIndex(f => f >= freq);
    return index === -1 ? frequencies.length - 1 : index;
}

/**
 * Calculates the necessary volume adjustment to level a speaker's response to a target
 * curve, based on an equal-loudness weighting derived from ISO 226.
 *
 * This function emphasizes differences in frequency regions where human hearing is more
 * sensitive, aiming for a perceptually balanced level match.
 *
 * @param {Float32Array} responseToAdjust - The measured magnitude response of the speaker (in dB).
 * @param {Float32Array} targetMagnitude - The target magnitude response (in dB).
 * @param {Float32Array} frequencies - The frequency points corresponding to the magnitude arrays.
 * @param {number} startFreq - The lower frequency bound for the calculation.
 * @param {number} endFreq - The upper frequency bound for the calculation.
 * @returns {number} The calculated volume offset (in dB) that should be applied.
 */
function calculateEqualLoudnessOffset(responseToAdjust, targetMagnitude, frequencies, startFreq, endFreq) {
    let weightedSumDiff = 0;
    let totalWeight = 0;
    const startIndex = freqToIndex(startFreq, frequencies);
    const endIndex = freqToIndex(endFreq, frequencies);

    // A reference loudness of 60 phon is used to determine the perceptual weights.
    // This is a typical level for moderate listening.
    const referenceLoudness = 60;
    const refSplAt1k = referenceLoudness; // At 1kHz, phon is equal to dB SPL by definition.

    for (let i = startIndex; i <= endIndex; i++) {
        // Ensure both magnitude values are valid numbers
        if (isFinite(responseToAdjust[i]) && isFinite(targetMagnitude[i])) {
            
            // Calculate the perceptual weight for the current frequency.
            // Frequencies where our hearing is more sensitive (e.g., 2-5kHz) will get
            // a higher weight because the equal-loudness contour is lower there.
            const splFor60Phon = ISO226_MODEL.getSpl(referenceLoudness, frequencies[i]);
            const weight = Math.pow(10, (refSplAt1k - splFor60Phon) / 10);
            
            // Calculate the difference between the target and the measured response.
            const diff = targetMagnitude[i] - responseToAdjust[i];
            
            // Apply the weight to the difference.
            weightedSumDiff += diff * weight;
            totalWeight += weight;
        }
    }

    // The final adjustment is the average of the weighted differences.
    return (totalWeight === 0) ? 0 : weightedSumDiff / totalWeight;
}