Speaker
Products
In The Lab
Forums
Newsletters
Join now
Best
Bluetooth Portable Loud Outdoor Home Bluetooth For Sound Brands Small Budget And Cheap JBL Bluetooth For Bass Parties Waterproof Bluetooth Shower Golf Smart Alexa Bluetooth Under $100 Bose Sonos
Tools
Compare Results Table Tool Review List Review Index Graph Review Pipeline Vote Custom Ratings
Popular
JBL Boombox 4 JBL Charge 6 Bose SoundLink Max Bose SoundLink Flex (2nd Gen) Ultimate Ears WONDERBOOM 4 JBL Authentics 500 JBL PartyBox Stage 320 JBL Flip 7 Anker Soundcore Boom 2 Plus JBL PartyBox 720 Anker Soundcore Select 4 Go JBL PartyBox Encore Essential 2 JBL PartyBox Club 120 JBL Boombox 3 Harman/Kardon Go + Play 3 Anker Soundcore Boom 2 Ultimate Ears EPICBOOM Harman/Kardon Onyx Studio 9 JBL PartyBox 520 Sony SRS-XB100
194 Speakers bought and tested
Supported by you via membership, and when you purchase through links on our site, we may earn an affiliate commission.
  1. Home
  2. Speaker
  3. Test Bench 1.0

Test Bench 1.0  
Changelog

 0
By Stephen Eldon KerrUpdated Jul 13, 2026 at 02:14 pm

See the previous 0.8 changelog.

We've updated our speaker sound testing methodology to Test Bench 1.0. This update changes how we measure and score Frequency Response Accuracy, Raw Frequency Response, Directivity, and Loudness. Because these tests changed significantly, Sound scores from Test Bench 1.0 aren't directly comparable with results from previous test benches.

What's Changed?

This update affects the following Sound tests:

Test Group Changes
Frequency Response Accuracy Updated measurements using sine sweeps, nearfield measurements, a 1 m microphone-array measurement, and a dedicated port measurement when applicable
Raw Frequency Response Updated graph using the same new measurement method, while still showing the uncompensated response
Directivity Renamed from Soundstage, measured with sine sweeps instead of pink noise, updated with a circular polar plot, and reweighted to focus more on off-axis consistency
Loudness Renamed from Dynamics, measured at 1 m using CTA-2034 weighted noise, updated with a new DRC calculation, and reweighted to focus more on maximum volume
Usage Scores Updated to reflect the renamed Sound tests: Soundstage is now Directivity, and Dynamics is now Loudness

Why Are We Making This Change?

Speaker measurements are especially sensitive to the room they're taken in. Our previous Frequency Response Accuracy method used averaged in-room measurements, which could be affected by room modes and other room effects, especially in the bass range. Test Bench 1.0 uses a new measurement method that better estimates the speaker's own frequency response while reducing the influence of the room.

We've also renamed and adjusted some Sound tests to better reflect what they measure. Soundstage is now Directivity, since the test focuses on off-axis consistency rather than subjective soundstage impressions. Dynamics is now Loudness, since the test focuses on how loud the speaker gets and how much its sound changes at maximum volume.

Frequency Response Accuracy

Frequency Response Accuracy now uses an updated measurement method. Instead of using pink noise and averaging 1 m and 2 m in-room measurements, we now measure the speaker's response using a sine sweep. We take three sets of measurements: nearfield measurements at seven positions, a 1 m measurement with a 15-microphone array, and a dedicated port measurement when applicable. This better estimates the speaker's own frequency response while reducing the influence of room effects, especially room modes in the bass range.

The in-room target hasn't changed, but the way we measure the speaker against that target has. This means a speaker's Frequency Response Accuracy score may change even if its general sound profile is similar.

An image of a graph showing our new Frequency Response Accuracy measurements for the Ultimate Ears HYPERBOOM speaker.
Our new Frequency Response Accuracy graph for the Ultimate Ears HYPERBOOM speaker.

We've also changed the score weighting. Std. Err. now has more influence because an even, balanced response is a major part of what makes a speaker sound accurate. Low-Frequency Extension is still important, but the new weighting avoids leaning too heavily on it, which can overly penalize smaller speakers that sound balanced but can't reproduce much deep bass.

Score Component Old Weighting New Weighting
Std. Err. 40% 50%
Low-Frequency Extension 55% 45%
High-Frequency Extension 5% 5%

Raw Frequency Response

Raw Frequency Response now uses the same updated measurement method as Frequency Response Accuracy. It still shows the speaker's uncompensated response, so the results aren't adjusted to match our in-room target. However, the target curve is still shown on the graph for reference. The new graph is based on the updated sine sweep, nearfield, and 1 m microphone-array capture process, making it more useful for comparing our results with external measurements.

An image showing our Raw Frequency Response measurement for the Ultimate Ears HYPERBOOM.
This shows our new Raw Frequency Response graph for the Ultimate Ears HYPERBOOM speaker. It's the same data as shown in our Frequency Response Accuracy graph, but it isn't compensated to our target. But you can still see our target overlaid as a dotted red line.

Directivity

Soundstage is now called Directivity. This name is a closer fit for the data in this box, which is built around Directivity Index and Stereo rather than subjective impressions of width or immersion. In practice, the test shows how much a speaker's sound changes as you move off-axis, like if you're sitting to the side or walking around it, and whether it keeps the left and right channels separate.

We've also updated how we capture and present these results. Directivity is now measured using the same sine sweep as Frequency Response Accuracy instead of pink noise. The graph has changed, too: instead of only showing a line graph, we now use a circular polar plot. This makes it easier to see how evenly the speaker projects sound around the room and where sound drops off as you move away from the front of the speaker.

An image showing our new Directivity graph for the Apple HomePod (2nd generation). The graph is almost entirely red because this 360-degree speaker sounds similar regardless of your listening angle.
Our new Directivity graph for the Apple HomePod (2nd generation). This graph shows how the sound of a speaker changes as you move horizontally around it. Since this speaker has a 360-degree design, it sounds very similar regardless of your listening angle.
An image showing our new Directivity graph for the JBL PartyBox 720. The graph is mostly shaded red in front of the forward-facing drivers, where you will hear the best sound. As you move away from the driver, the graph is colored blue and green to show mid-range and treble drop off.
Our new Directivity graph for the JBL PartyBox 720. Since this is a forward-facing, PA-style speaker, its sound changes more noticeably as you move horizontally off-axis from the drivers. In particular, mid-range and treble frequencies sound quieter.

Directivity Index now carries more weight because this box is mainly about how consistent a speaker sounds off-axis. But stereo still counts, since separate left and right channels affect reproduction accuracy, especially for sounds like hard-panned guitars and spatial effects like reverb.

Score Component Old Weighting New Weighting
Directivity Index 75% 85%
Stereo 25% 15%

Loudness

We've renamed Dynamics to Loudness, since this test focuses on how loud the speaker gets and how much its sound changes at maximum volume.

We now measure Loudness at 1 m using CTA-2034 weighted noise. DRC at Max Volume also compares the speaker's response at maximum volume against its response at 85 dB SPL instead of 76 dB SPL. DRC at Max Volume is now calculated from the speaker's Low-Frequency Extension to 20 kHz, rather than across the full 20 Hz to 20 kHz range. This helps avoid penalizing speakers for compression in frequencies they don't reproduce.

Maximum volume now makes up most of the Loudness score, so the result is easier to read at a glance: higher-scoring speakers get louder. But DRC at Max Volume still counts, since a good speaker should stay clean and controlled when pushed near its limits.

Score Component Old Weighting New Weighting
SPL @ Max Volume 60% 90%
DRC @ Max Volume 40% 10%

Usage Scores

We've updated the usage score weightings to reflect the renamed Sound tests: Soundstage has been replaced by Directivity, and Dynamics by Loudness. Since the underlying Sound tests and score weightings have also changed, some usage scores may change even if a speaker's general performance is similar.

What Do You Think?

Our process is iterative, and your feedback is instrumental to our test bench development. If you have comments or questions about the above changes, please reach out to us in the comments, in our Discord server, or via email.

We are retesting popular models first. So far, the test results for the following models have been converted to the new testing methodology. However, the text might be inconsistent with the new results.

    92 Speakers Planned To Be Updated

    We are also planning to retest the following products over the course of the next few weeks:

    We make money directly from you via Membership (which gives you access to extra features) and when you buy a product after clicking a link to a retailer. Learn more about how we work.
    We buy our own products, just like you, mostly from the United States. When a product is discontinued or no longer popular, we sell the review units locally. See what products we currently have for sale.
    © 2026 9298-5266 Quebec Inc. All Rights Reserved.