Our Imaging Score And Tests

Headphones

Group delay is the time it takes for the amplitude of each frequency to reach its maximum. Group delay looks for single-channel frequency-dependent time distortions created by the headphones. In other words, it checks to see if a portion of the frequencies in the response of the headphone arrives later than the rest. Group delay differs from phase mismatch by being a monaural quality; it can be perceived even with one ear. But phase mismatch, by definition, is a stereo quality.

For our group delay calculation, we've implemented a weighting filter based on several studies done on the audibility threshold of group delay on headphones at different frequencies to make our results more perceptually relevant. Since the research was limited to the 500Hz-8kHz range, we extrapolated the results for the bass and high-treble regions. Below is our current audibility threshold for headphone group delay in milliseconds.

Usually, headphones have higher group delays in the bass range compared to the higher frequencies. For example, it may take 25ms for a headphone to produce a 40Hz tone at 90dB SPL, while it can produce the same amplitude at 1kHz in less than 2ms. A high amount of group delay in the bass range results in a slow and loose bass. A high amount of group delay in the treble range negatively affects the transparency of the sound. Some Bluetooth headphones have a spike in group delay around 20kHz, which can be caused by different things, including signal interference. However, this won't be audible to most people.

This test measures the amount of difference between the left and right driver's phase response. If you're not familiar with phase, we'll need to go back to the basics of sound. One way of representing sound is by using sinusoidal waves. The height of a wave tells us its amplitude (or loudness), while the length of the wave indicates time. Since soundwaves have a repetitive form, we want to focus on a wave's cycle or wavelength.

Phase is the location of a point within a repetitive sound's wave cycle, which we can express in angles, starting from 0° until 360°. For example, if we're looking at max amplitude, we want to look at phase at 90°, as this is the highest point within the sound wave's cycle. However, this on its own doesn't tell us much. Phase is most useful when we can compare the differences in phase between more than one signal.

 

 

Several soundwave cycles.¹

 Two sinusoidal waves with the same frequency but have phase difference or phase mismatch².

Stereo headphones use a left and right driver to produce coherent sound as our brains process both signals simultaneously. Assuming you're using a consistent tone, each time the left driver's sound wave reaches its peak amplitude at 90°, the right driver's sound wave should also reach its peak amplitude at 90°. When both waves are matched in phase, their amplitude increases. In headphones, a well-matched phase response indicates a stable stereo image. However, a bad mismatch in phase response can point to inaccuracies in the stereo image reproduction at certain frequencies. At worst, it can push objects to the sides and leave a hole in the middle of the stereo image. 

Our graphs reflect when you'll hear phase mismatch in different frequencies. When the response crosses above the audibility threshold, the mismatch may be audible. For our phase mismatch calculation, we calculate the standard error of one channel phase response against the other, but we have implemented a weighting filter based on the research done on the audibility threshold of phase mismatch on headphones at different frequencies to make our results more perceptually relevant. Since the research was limited to the 63Hz-8kHz range, we extrapolated the results for the sub-bass and high-treble regions. Below is our current audibility threshold for headphone phase mismatch in degrees.

That said, the amount of deviation from 0° and the frequency range of the peak can impact whether or not you hear the mismatch. For example, the Focal Celestee have a peak at just under 16KHz, or in the high-treble range. However, this can be very hard to hear, especially with real-life content. This is because the higher the frequency, the harder it is to sync each wave cycle as the degree between them is becoming smaller and smaller. It's also harder for us to detect these mismatches with our ears, and we lose sensitivity to this range as we age.

In another example, the JBL Tune 510BT Wireless have a more prolonged peak between 200Hz and 900Hz, or between the high-bass and mid-mid. This peak indicates that the amplitude of these sounds won't be representative of the amplitude desired by the audio you're listening to, and they sound quieter than intended. You can hear it in real-life content, but it may be more obvious to some users than others. 

A lot of phase mismatch can look like the Anker Soundcore Life Q30 Wireless' graph. There are a lot of peaks, which indicates that objects like voices skew to one side of the stereo image. This also means that the phase mismatch is noticeable in real-life content. When listening to your favorite tracks, audio is skewed to either the left or right driver at different frequencies, making for a harsh and fatiguing listening experience.

This isn't a design test, but since it's essentially a driver matching test, it could also be considered a marker for manufacturing tolerance and ergonomics.

Weighted amplitude mismatch assesses a model's left and right driver's balance with amplitude or volume. When headphone drivers are matched in amplitude, they can reproduce objects in the intended location within your mix. Conversely, high amplitude mismatch can create unwanted shifts in the stereo image and potentially lose or exaggerate details. For example, if a pair of in-ears have 3dB of amplitude mismatch favoring the left channel, the stereo image's center (and the objects within your mix) will shift to the left.

This isn't a design test, but since it's essentially a driver matching test, it could also be considered a marker for manufacturing tolerance and ergonomics.

Our sound graphs are level-matched at 90dB SPL and 100dB SPL so that test results are consistent and comparable across reviews. However, these graphs don't show the actual measured amplitude mismatch between the L/R drivers of the headphones. Instead, we report this measurement in the Imaging box under 'Weighted Amplitude Mismatch'.

This test assesses the amount of difference or standard deviation between the frequency response produced separately by a unit's left and right drivers. This mismatch results in an uneven stereo image where some frequencies are skewed to the left or right instead of being centered. Frequency response mismatch is similar to amplitude response mismatch, as in they look for the same kind of error, but frequency mismatch does it per frequency, as opposed to amplitude mismatch, which does it per channel.

That this isn't a design test. Since this is essentially a driver matching test, however, it could be considered a marker for manufacturing tolerance and ergonomics.

For example, the Samsung Galaxy Buds Live Truly Wireless display minor frequency mismatch. Their left driver's bass range is more emphasized and boomy than their right driver, which could offset sounds like thump and rumble in EDM tracks and skew them to the left. Conversely, the left driver's mid-range is less emphasized than the right driver, offsetting voices and lead instruments to the right.