TestHiFi intro and motivation
Non-technically interested friends used to ask why we do such a thing as the TestHiFi app.
I could have answered with the words of another friend, who, when asked “why” about his super-elaborate tube amplifiers, just answered “because I can”. But that’s not true. We created TestHiFi for those less technically interested, but music loving people.
It’s been a while since we started to work on it. But if I remember correctly, Willi wanted to create a simple app that can tell within minutes if an audio system performs close to HiFi or less than that. Of course I instantly assured this is absolutely impossible with just a mobile phone.
His motivation stems from the fact that after the introduction of data-reduced MP3 audio, the next step down manifested itself in form of bluetooth systems. Many of these are really far away from reproducing anything like a real HiFi system should. Easy separation of good from bad might move consumers back towards better quality?
My own motivation adds the hope that if more people use higher quality gear, music would more often be enjoyed as an inspiring primary event instead of monotonous background beats.
As you can see, we faced the challenge. I had my own agenda, hoping we would arrive at providing more than the targeted “traffic light” response from the app … one day. To replace the simple FFT analyzer apps I use to check systems on the fly.
But even the original idea – just a simple “not good” – “ok” – “close to hifi” test result – kept us busy for months.
There are countless ways to make more or less meaningful measurements of audio systems.
Let us take a closer look at our two main ingredients.
For decades now the most prominent measured feature of an audio system is “frequency response”. This indicates if the system reproduces the full frequency range of sounds recorded, without altering the timbre. Unfortunately, measuring this within domestic boundaries always includes strong room influence. Hence, the same system in different rooms measures differently. Not good to judge if that system measured at the dealers or friends place will perform better than ours at home.
Our solution uses two strategies to reduce the problem:
Firstly, measure the system close-up, at about 50 cm distance. The closer we move towards the sound source, the lower the relative influence of room reflections on the measured frequency response. But there is a limit. For several reasons, we should measure the system at a distance larger than the sound source’s size. In short, this is about sensing the contributions of several radiating parts equally.
To improve the result, we take additional measurements at different positions. The results are interpreted by our algorithm in ways to exclude both some more room influence and phase problems due to different distances between the microphone and different loudspeaker elements. (No, it’s not a simple average of the individual results. This is what I miss from FFT apps. And why I am happy that we added listening position measurements and graphic output – which makes it useful for me.)
For example, a 2 way loudspeaker will measure very different from above, on-axis or below tweeter height. Same to systems with additional woofers etc.
The result is a quick and condensed version of more elaborate strategies where many measurement positions are used to “average” the response and find out what is produced by the sound system and what is contributed by the room.
Over the past decades loudspeakers became smaller and smaller. As explained in another blog ( https://testhifi.com/2020/01/17/woofer-size/ ), to produce low frequencies efficiently, a diaphragm must be large. Therefore, miniature speakers struggle to produce real bass. Assisted by digital filters, they may provide acceptable results at low levels, but distort heavily at higher levels. To prevent the small loudspeaker from overload, sophisticated algorithms control low frequency levels. Some shift their energy one octave higher than the original sound. This cheats our ears to perceive the original frequency, but puts less stress on loudspeaker and amplifier. Many more sophisticated methods to squeeze louder bass from small devices were created by mobile phone technology.
Small bluetooth systems not only have limited size speakers on board, but also limited amplification power and battery capacity. Small speakers work with low efficiency, consuming a lot of power to play louder than background levels. Add the need for filters to enhance the meager low end, which again demands lots of amplification power. This is the reason why such systems run out of steam when asked to play louder, with high distortion.
To check if the device under test can reproduce reasonable levels with acceptable quality, we measure “distortion”. Sounds which are not part of the music (or test signal). While there is a lot of discussion about how distortion is perceived, at those levels we look at they are certainly limiting the audio quality.
While testing app prototypes, we measured similar bluetooth systems with very different distortion profiles. Some otherwise ok sounding distorted drum beats and our test signal brutally. And at low battery levels, even good performers lost a lot of quality.
To measure distortion, the reproduced test signal is compared with the level of artificially added sounds produced by the device under test.
Again, room contributions can distort these relations by enhancement and cancellation of frequencies. (See our blog about comb filtering : https://testhifi.com/2019/01/14/acoustical-interference-and-comb-filtering/ )
Hence, the 3-position measurement close to the system with some “intelligent averaging” makes sense again.
Of course, the little app cannot replace sophisticated laboratory equipment. But it provides a fast and simple way to measure basic performance parameters, using a device that most of us already carry around. With the recently added listening position test and graphic output, it does all I need for quick evaluations. What I still miss, is a simple and reliable method to connect the mobile phone to the audio system. 3.5mm connectors fit notoriously loose, cables need to be long, and wireless options have their own problems. Let’s drink some tea and see what the future offers.