As a hobby, I develop and build speakers - along with their crossovers. Thus, I have access to calibrated audio measurement equipment that I let loose on the Framework Laptop speakers.
Technical detour
My measurements are calibrated, meaning that my microphone and sound card were tested against a known flat reference in a testing laboratory. I only measure a single laptop speaker at a time in a so-called near field measurement, where I place my microphone in very close proximity to the speaker. This neglects the influence of any baffle to which the speaker is mounted. I decided this procedure to be valid, as I assume this “baffle” to be close to infinite in size (the table that the laptop sits on). Surely, this is not ideal, but it is quick and easy.
Measurement results
See below the frequency response before (green) and after (brown) applying my filters. Most notably, there is a large hump at around 500 Hz that makes the sound very muddy and unbalanced.
Of course, the speakers are still on the thinner side. The lower cutoff frequency is as high as 370 Hz (-3 dB). However, I find it much easier to listen to with the added filtering, especially for voices.
After installing it and activating it for only the speaker output of your machine, create this configuration file in the config folder of Equalizer APO
Filter: ON PK Fc 485 Hz Gain -5 dB Q 0.7
Filter: ON PK Fc 3500 Hz Gain 3 dB Q 3
Filter: ON PK Fc 5000 Hz Gain -6 dB Q 2
Filter: ON PK Fc 8000 Hz Gain -3 dB Q 2
Filter: ON PK Fc 10700 Hz Gain 12 dB Q 3
and call it something like framework_laptop.txt. Then, open the Configuration Editor that was installed alongside Equalizer APO and include this configuration file. In the end, it looks like so:
Disable other Windows sound enhancements
You should not enable any other sound enhancemehts. The advanced audio settings page should look like exactly this, with no boxes ticked.
Yes, that is right! You can change settings and toggle it on and off.
There is also the option to limit the filter to a certain output device, in case you installed APO for all of your devices (or at least multiple). You can add the filter option like this (sorry, it’s german):
It says “Control” > “Device (choose device)”
Thanks for this nice guide and good idea!
I tried this out on Linux with the EasyEffects Tool and it seems to work. Sound seems clearer with the EQ applied but I did not test with the laptop on a table yet.
Be aware that you need to add an automatic preset to load the EQ profile while using the laptop speakers and another one without the EQ for the headphone jack which is only possible with actual headphones connected.
Recently, I also tested extending the bass into lower octaves (e.g. 200 Hz), but that was not successful. The amplifier is too weak and volume drops on bass notes. So, we are stuck with how it is.
Ah yes, a very similar approach. While the filter coefficients are different, the outcome is probably quite similar.
I still want to do a measurement in a setup like that, where the laptop is on a table and the microphone over top. Unfortunately, in quick testing, the frequency response depends a lot on the position of the microphone, so I have to perform some averaging over different listening positions.
Usually, integrated Class-D amplifiers, as used in this laptop or other mobile devices, include a digital signal processor (DSP) component. That has become very common, even in the cheapest circuits.
Filter coefficients can be loaded at production time, and the speakers are driven with corrected inputs, without the need for any software.
Sure, I will provide new measurements when I make them!
The speakers can certainly be tuned to sound much better to most of the listeners.
Please note that the settings I posted are not subjective, they only remove the defects in frequency response from the speaker outputs. Filter parameters were optimized such that the resulting frequency response is flat. Thus, the speakers represent the sound as it was intended to be listened to (apart from the missing lower and higher octaves, which the physical speaker is not capable of reproducing).
If the frequency response is not flat, the speaker introduces some characteristic “sound” which a listener might like, but that is not fully representative of the source material anymore.
However, the deviations that the Laptop speakers show (10 dB at crucial frequencies) are not acceptable “sounding” anymore, in my opinion. The speakers are just not engineered very well.
@elagil : Thanks for your work. I’ve had my Framework 12th gen for 2 weeks and that bump in the mids was quite annoying on speech content, especially as i am a sound tech/musician
Yes, I have only the base model and don’t have measurements of the updated ones. Presumably, they just have a lower impedance coil for higher efficiency. Maybe they measure in a similar way.
