Real World Angry Box Measurements

Last year, I promised to take more detailed distortion measurements of the Angry Box, our true point-source, ultra-compact loudspeaker. It's now well into 2025, and other Tantrum-related priorities took the lead: bootstrapping our current workshop facility in Surrey, UK; building out our dealer network; developing exciting new products (coming soon!); and, of course, the delightful but overwhelming fatigue of raising a child under two while starting a frickin' speaker company. Sorry it's taken me this long to get around to this write-up, but I hope you enjoy it regardless!

So, lets get into it then...

The Measurement Process

All the measurements were calibrated using pink noise to 88 dB SPL at 1 meter and taken at our workshop in Hook, UK. The workshop is inside an old grain silo, a circular metal structure about 3.5 meters in diameter, with a cool trick: if you stand precisely in the center and talk, you can hear your voice coming from behind you...

While the space is otherwise fairly well behaved, perfectly suitable for relative measurements against a known good reference by analyzing the difference signal, it does have limitations we need to talk about when it comes to capturing absolute loudspeaker data of the kind we are taking today. I am, at least in part, writing this to challenge the notion that loudspeaker measurements are inherently difficult, and to show that you don't always need an anechoic chamber to collect useful data. And maybe it will inspire you to measure your current speakers.

To reduce the room's influence as much as practical, I did a few things. First, by placing the speaker up on the bench and away from the wall (and away from the center of the room), I was able to reduce boundary loading, which is generally desirable for free-field measurements. That said, boundary loading is actually something I feel benefits the Angry Box. It gives the speaker a nice low-end lift, rather than the low-mid muddiness you often get when a speaker is placed too close to a wall.

Secondly, after level calibration, I moved the microphone much closer to the speaker, to around 5 to 10 cm away (picture below), to minimize the room's reverberation in the measurements. I also reduced the preamp gain to avoid clipping, while keeping the playback volume consistent.

This helps us achieve workable measurements due to the way sound propagates according to the inverse square law. As you double the distance from a sound source, the sound pressure level decreases by 6dB. Conversely, halving the distance increases the sound pressure level by 6dB. Moving from 1 meter to 10 cm results in a 20dB increase in sound pressure, while the room's impact is reduced by approximately 30dB due to the closer measurement position. While nowhere near the level of anechoic chamber measurements, this setup is sufficient to reveal speaker issues that exceed the noise floor of a typical room, meaning any problems not detected here are almost certainly inaudible in real-world listening environments. Seasoned professionals may still raise an eyebrow at this, more established brands typically wouldn't publish such data. However, at this stage in the company, I believe transparency is more valuable.

The Silo

We cannot eliminate the influence of the room with this setup in the same way that a good anechoic environment or a Klippel measurement system can, so we will be limited in the time domain by our reverberant noise floor. We are at least able to give ourselves enough headroom in the measurements to gather some useful time-domain and THD data. Perhaps one day, I'll revisit these measurements when I have access to a more ideal facility.

For now, rather than hiding the imperfections, I think it's better to share what I've captured for everyone to see...

The Measurements

As for my measurement equipment, it's far from the most advanced kit available, but I'd actually argue that this isn't all that important when measuring a loudspeaker either. While I'm using a fairly good measurement microphone (the Earthworks M30), the £30 Behringer mic can do just as good a job.

And while at home I have an RME interface with a much lower distortion and noise floor than the interface I used today, Jamie Anderson, founder of Rational Acoustics (makers of the popular SMAART software for acoustic analysis), once pointed out during a SMAART training I attended that the distortion and non-linearities of a typical loudspeaker far exceed the noise floor of even the cheapest audio interfaces and measurement mics, meaning your equipment's noise and distortion are effectively masked by the speaker's own distortion.

If we were measuring electrical circuits, however, that would be a different story. For the sake of brevity, though, I've also taken measurements of the audio interface I had on hand, the Focusrite Scarlett 2i2 🤮, so you can understand our measurement baseline.

Scarlet 2i2 Frequency Plot and Distortion.

I ended up using REW (Room EQ Wizard) for all of these measurements. I'm fairly new to REW, as I come from a live sound background and have spent most of my time measuring speakers with SMAART. While SMAART is fantastic for loudspeaker tuning and calibration with its real-time capabilities, it's not as well-suited for full-spectrum harmonic distortion analysis or for generating shareable graphs for reporting.

Angry Box Frequency Response Plot

This is a frequency plot of both modes on the Angry Box, taken with REW. The extended range mode is shown in green, and the mid-focus mode is in red. Nice! It still vaguely resembles the measurement I took in my home office with SMAART before launching the product over a year ago, which became the plot on the back of the speaker. This is good to see.

Believe it or not, there's no smoothing on the above graph—just an average of 4-5 measurements taken within 10 cm in front of the loudspeaker. It's smoother than I'm used to seeing with raw traces in SMAART. I suspect the difference comes from SMAART averaging windowed pink noise FFTs, while REW uses a sine sweep.

Angry Box Harmonc Distortion Plot

Left Extended Range | Right Mid-Focus

Above are the THD plots, with extended range on the left and focus mode on the right. For those interested, I've placed them on the same scale so you can compare them to the distortion plot of the 2i2 above.

As you can see, the distortion remains very low across the midband, in line with the design intent of the speaker. It stays under 0.5% for the entire usable frequency range, with some areas as low as 0.1%. The distortion rises in the low end, which is inevitable with a small speaker, but it still stays under 1% in the mid-focus mode. Overall, not bad at all.

Angry Box Group Delay Plot

Another primary design goal was to keep group delay as low as possible across the usable frequency range. For those who may not be familiar with the term, group delay refers to the time it takes for a signal to pass through a system at each frequency. The lower the group delay, the better, as it ensures the frequencies are more closely time-aligned, allowing all of them to reach the listener simultaneously. A characteristic naturally inherent to a true point-source speaker.

Imagine the sound of a snare drum: the crack of the snare and the thud of the drumhead must hit at the same time for a natural, cohesive sound. In a speaker system, inconsistent group delay misaligns these components, making it difficult to make accurate EQ decisions and, at worst, causing the snare to sound muddled or unnatural.

I feel the impact of group delay is often ignored in loudspeaker design (looking at you, ported and passive radiator speakers). Many of us have likely experienced electronic dance music on a large ported system. The lowest frequencies, reinforced by the port, are delayed by a whole cycle due to how the port works. This is especially noticeable on systems with very low tuning frequencies, where the delay creates a syncopated 'pt-uh' effect, which can sometimes sound quite good and even improve our perception of the bass separation. However, it's not ideal for studio playback systems where accuracy and time alignment are key priorities.

It's possible to reduce apparent group delay with DSP (Digital Signal Processing), often using a type of filter called an FIR filter to delay frequencies that arrive early. Many loudspeaker manufacturers apply this technique to correct phase alignment in 2- or 3-way systems. However, it comes with trade-offs—most notably increased system latency, since the entire signal needs to be delayed by at least one full cycle of the lowest frequency being corrected. It can also introduce pre-ringing on transients, which, in my experience, can negatively affect stereo imaging. Personally, I prefer to get things right at the source with a point-source design, careful driver and amp selection, and then use DSP for that final 5%, much like a good mastering engineer would. As a bonus, this approach keeps the total latency of the Angry Box under 1 ms. Win.

Angry Box Waterfall Plot

Finally, we have the waterfall plots above. This is a first for me with the Angry Box, since SMAART—while great for real-time RTA—doesn't let you capture waterfall snapshots like REW does (at least not as far as I'm aware). It's super neat to see! REW really is an impressive piece of free software.

A waterfall plot gives another perspective on a speaker's time-domain behavior—especially its resonances. If a speaker is ringing, it'll stick out like a sore thumb here. If this is your first time seeing one, go Google ‘ported loudspeaker waterfall plot' and you'll see exactly what I mean.

Interestingly, the plots actually resemble those of the Yamaha NS10M. This makes sense, as both are relatively small sealed-box loudspeakers.

Yamaha NS10m

I hope you enjoyed the read and learned a bit about what we're up to at Tantrum. While the setup isn't textbook perfect, I've done my best to control what I could and be transparent about what I couldn't. Hopefully, this gives you a fair glimpse into how the Angry Box performs and the thinking behind its design. If you think i have made any mistakes please let me know. You have my email.

It's easy to get lost in specs and graphs, but at the end of the day, you've got to hear a speaker to truly understand what it's all about.. What I've aimed for here is something honest, tight, and time-aligned, something that punches above its weight and helps audio professionals like you deliver better results.

Thanks for sticking with me if you made it to the end. If you'd like to hear when we release new things, feel free to subscribe to the email list below. I'll do my best to keep you updated in the least boring way possible. We might even have something new coming soon 😉

Thanks again for reading, and more to come! - Tom Turner, Founder and Designer of The Angry Box.