What is Soft Clipping
Discover the art of soft clipping in audio production: uncovering the balance between clarity and character.
I’m sure you’ve heard about frequency response when shopping for headphones or speakers. It is one of the more significant specs many audiophiles and enthusiasts talk about when referring to the overall sound quality of audio devices.
So what’s the big deal about frequency response? Is it a reliable metric for figuring out if a headphone/speaker has premium sound?
Simply put, the frequency response of a headphone or speaker helps users get an idea about how loud and how accurately the bass, mids, and highs are going to be translated by the drivers. Most often, this ability is measured and visualized by a frequency response graph showing the frequency with its corresponding loudness (in dB). Comparing this response against other devices or a predetermined target curve can help users figure out if these headphones will suit their needs.
However, frequency response alone does not qualify a headphone or speaker as being “better.” There are a lot of other factors involved.
Let’s talk more about frequency response and how it would help you get a better idea about the performance of an audio device.
The frequency response is a metric used to denote the overall sound profile of a headphone or speaker system.
A headphone’s frequency response can be visualized by plotting the frequencies (that the drivers are capable of producing), along with their corresponding amplitude using a bode plot.
In this manner, we can measure the loudness (in dB) of each frequency and group that information in a graph; hence, we can see the entire range of frequency bands that the drivers can give out along with the corresponding loudness of each frequency.
As long as the compensated frequency response does not stray too far off from the target curve, the audio is going to come out clean and accurate.
The quality of headphones (or speakers) depends on the drivers. Therefore, if the headphone drivers can produce a sound that stays true to the input signal, it’s widely accepted that the headphones have good drivers that deliver high-quality audio.
However, judging the quality of these headphones and speakers by listening through them is not the most accurate since everybody has a different experience; hence, it is very subjective.
Therefore, one of the more objective and reliable methods to determine the quality of the audio signals (reproduced by a headphone or speaker) is to consider the frequency response graph.
If headphones (which is an output device) reproduce all the promised frequencies in a clear and loud enough magnitude (or amplitude), that driver is considered capable of recreating accurate and/or enhanced sound.
In other words, it is universally accepted as a high-end audio output device with great sound.
Understanding frequency response graphs are pretty simple as long as you know about the bass, mids, and treble and where they sit in the audible frequency range. (I’ll explain more about the audible frequencies and the three frequency categories in the FAQ section.)
The graph has two axes, the one going from left to right (X-axis) shows the frequencies in Hertz (Hz) and these usually range from 20Hz to 20,000Hz. So it is not a linear scale but a logarithmic one. (Also called a bode plot.)
Going from the top to the bottom, we have the loudness, also referred to as amplitude or magnitude (pop pop). This unit is measured in dB (decibels).
To give you an idea about decibel values: 30dB is considered the equivalent of whispering while 90dB is similar to that of a lawnmower. Also, most frequency response curves use 90dB as a reference point.
(BTW, Here’s a list of headphones that you can wear while mowing the lawn.)
There are two types of frequency response curves, the raw frequency response, and the compensated frequency response.
The compensated response curve may not be the most accurate. However, it is easier to understand because it takes the 20Hz-20kHz range into account, and it is averaged to minimize errors caused by loudness settings and stereo imbalances.
Also, it is easier to make comparisons because it is compatible with the Harman Target response (The Harman target frequency response curve is widely accepted as the industry standard for headphones.)
In most compensated graphs, you’ll most likely find the target curve as a straight line at 90dB. This flat frequency response shows the ideal values for accurate audio reproduction. However, such a perfectly flat frequency response is difficult to achieve in real life, and it is not guaranteed to sound the same for everyone.
Therefore, you will always find deviations from this target response. What’s noteworthy is the amplitude of this deviation compared to the target curve and the other frequency ranges.
For example, most casual-use closed-back headphones (such as the Sony WH-1000XM3) will have an emphasized bass response. It is done purposefully to make the bass sound boomier. However, it doesn’t deviate too much from the target response and has a 3dB or 5dB deviation.
Generally, as long as the frequency response curve doesn’t stray too far from the target line, you will have a somewhat accurate response. On the other hand, if you are not using these headphones or speakers as studio devices, a perfectly flat frequency response may not offer a more immersive listening experience. (Sometimes, you need more bass to feel the music.)
Let’s see the frequency response of the most well-known TWS earbuds, Apple’s AirPods Pro.
From the sound profile graph of rtings.com, we can see that the bass response is slightly lower than the target curve throughout the entire bass range. (With a standard deviation of 2dB or 2.5dB.) This small deviation is quite acceptable, although some experienced users will notice the lack of bass sounds.
At the mid-range, we have a slight peak of 94dB at 507Hz, but it dips down and tries to settle with the target value resulting in mid-range audio (such as vocals and lead instruments) that are clear and accurate.
Finally, at the high frequencies (also known as the treble), we can see a dip at the 5.5kHz frequency. This implies that the high-end frequencies, such as cymbals will sound a little weak.
Now, if we take a look at the JBL PartyBox 710’s frequency response curve (one of the most premium Bluetooth party speakers), we can see that it has a lot of fluctuations.
This is quite a common phenomenon in speakers, mainly due to their size and reflections. Reflected sound waves superpose on the initial signal, causing the resulting frequency to increase or decrease in loudness. We cannot get rid of this behavior because not everybody can host a party in a reflection-free anechoic chamber.
Despite this, it is very easy to read the frequency response of speakers. You can either refer to the linear regression line (which tries to “average” the loudness values over the entire range) or draw a line through the resulting peaks. Nevertheless, we can see that the curve doesn’t stray too far from the flat target line.
For the JBL PartyBox 710, we can see that the overall bass (at the lower frequencies) is slightly more overpowered. (Despite having a sharp dip in the mid-bass.) This results in the speakers giving a punchier and boomier bass, especially at the low frequencies.
However, at the mids, you can see that despite the oscillations, they are stable with smaller dips and peaks. The result is an overall neutral frequency response where the reproduced sound is accurate and stays true to the initial audio signal.
Also, you can easily observe that the highs (treble) are considerably lower, causing high-pitched audio frequencies to come out a little weaker on the speaker’s end.
Measuring the frequency responses of a headphone or speaker is quite simple. However, you need professional noise isolation equipment, a measurement microphone, dummy heads, and an echo-free critical listening room.
The raw response is measured by feeding pure tone frequency signals (with the same amplitude) and recording the headphone’s output. After that, the measured frequency response is plotted on a graph.
Multiple measurements are taken in this way by readjusting the headphones on the outer ear of the dummy. Multiple raw frequency response measurements make it easier to compensate for conditions such as human hearing and different loudness levels.
Then, the final frequency response plot is created by averaging these graphs and narrowing it down to the specified range of 20Hz to 20kHz. (The audible frequency response range.)
As I mentioned in the introduction, the headphone frequency response alone is not enough to gauge the sound quality of your desired device. There are several other specs you need to consider.
Frequency Range: The frequency range defines the minimum and maximum frequencies the headphones/speakers can reproduce. It doesn’t give us a direct indication of the headphone frequency response. For audio output devices, a narrow range of frequencies is definitely not a good thing, so ensure the headphone’s frequency response is 20Hz-20kHz.
Sensitivity: The sensitivity of headphones and speakers refer to the loudness (also called the sound pressure level) measured at 1mW or 1W for headphones and speakers, respectively. On average, a headphone output range of 90dB SPL/mW to 105dB SPL/mW is usually considered loud enough without noticeable distortion. (The same values apply to speakers with W instead of mW.)
Impedance: Also referred to as electrical resistance, the impedance acts as the load for the audio signal. Therefore, to “drive” this load and generate quality audio, you need more power. Usually, high-end headphones have higher impedances, and they need to be driven by amplifiers. Yet, they have better quality audio. (Because they use less current to drive the same amount of power.)
A headphone’s frequency response curve can be a great indication of how well the drivers can recreate certain frequencies. A good frequency response curve will always have an equal amplitude throughout the entire range of frequencies and result in a balanced sound that is clear and accurate.
Unfortunately, headphones that are tuned to have perfect frequency responses are hard to come by, and most often, they are not guaranteed to provide an enjoyable listening experience.