top of page

Thoughts on the Muse S Brain Sensing Headband

Updated: Jun 14, 2022

I’d bought the Muse headband quite a while ago now and have used both the standard Muse app that comes with the band and also the great 3rd party app, Mind Monitor.

There’s 2 different types of the headband from the date of publishing this post: a more rigid framed band, and a flexible fabric one (as in the image above) that is intended for better comfort and ability to be worn during sleep more easily.

The actual hardware and most of the app capability works across both the Muse 2 and Muse S but the Muse S supports additional sleep related features such as guided sleep using soundscapes and audio guides.

I was particularly interested in some of the new sleep analysis features released for the app, which looks to gauge your quality of deep sleep and sleep stage breakdown. A couple of years ago there was a device developed by Dreem that looked very promising also but it's no longer available and the company have stopped development in this space so the Muse S has virtually no competition at this level.

The other reason why I purchased the product was from the knowledge that raw and processed EEG data can be exported quite easily from the band via the 3rd party application, Mind Monitor, which has been very well received on both Android and IOS app stores. I’ve worked with HR data and devices before but have always wanted to see how that related to my EEG signals, in particular for meditation and sleep. EEG and brain signals provide a whole new set of metrics to work with.

The first thing I tried was the native app made by Muse. Setup is fairly straightforward but one thing I did find a bit irritating on the sensor check aspect is getting good enough contact on the side sensors that go above your ear – sometimes it takes a while. You do need to make sure you put a small amount of water on the sensors before putting it on and check you’re getting enough signal across all of them.


For meditation, the MUSE app provides various modes and guides related to things like breathworks, heart and body meditation and these are all great quality. The metrics you get back after a meditation don’t really relate directly back to the EEG and HR data, rather they are processed to provide MUSE specific metrics that I guess are supposed to be easier for a consumer to make sense of with things like: Muse points, ‘Birds’, and Recoveries. Recoveries relate to how well you recover from a distraction, Muse Points are constantly earned through meditation each second and Birds are given when you’re at a meditative state for long enough time. Also, rather than Alpha, Beta and Delta waves, you're given a line graph that tracks your state from 'calm' to 'active', again as it's easier to digest at a glance:


For sleep, you do need to pay particular attention to wearing it with the right amount of tension so that it stays on overnight – I have to say this device is as comfortable as these could be, it’s light and not in the way all that much. That said I do move around a lot and therefor this does end up moving the band to an extent. The Dreem headband seems like an interesting design that might have worked better to counter head movement. I prefer the metrics and information you get for sleep vs meditation, again you’re given a generic score but also a hypnogram, a graph charting deep sleep also info related to head/body movement and your HR. This to me is a great level of information – one problem I have though is that it rarely manages to work across a whole nights sleep, there’s usually always some blank spots where the headband loses enough signal to stop it tracking brain signals. The battery life is good enough, it will run on a full charge for around 10 hours and I think that’s fine given the amount of data it’s taking in and the processing from the various sensors.

I have mixed feelings about the native app – I do think it definitely makes sense for most consumers so it’s not a criticism when I say my preference is still for more detailed info and metrics that I can use myself and process offline. And in fairness they do (or did) provide access to the SDK in order to make 3rd party apps and access the data directly. Although the SDK is seemingly now unavailable, a very comprehensive app was developed that provides all the data you would want from the band:

I tried the app using the default settings and a DropBox account so that exports can be uploaded for later analysis and was impressed by how well it works. It’s not the cheapest app at all but I’d say it’s worth it given it’s a one-off payment and based on how well it functions with reliability across all the measurements you can record. You can also view charts of the data directly via the mind monitor website – it’s a JavaScript based charting tool and so the data is never sent over the web, it’s all done in the browser locally.

This is an example from one sleep session below. I think the browser graph tool is a bit limited for this situation where you have hours of data because it doesn't average it out long enough to view as a trend so looks quite spikey but for meditation and short durations this view is probably all you need:

There is also an Excel macro available, which I think scales up a bit better for large datasets:

I've actually set things up in my own way using Python that take the mind-monitor csv export and then outputs relative brainwaves in a smaller csv file with just the columns I need. You can see an example from a different night's sleep here with also a bar chart on the same chart area for the gyro sensor that helps identify wakefullness and light sleep:

One view that looked really interesting to me in the Mind Monitor app was the Spectrogram, this seemed to me to be the clearest way to see brain wave patterns/changes as well as bad signals and noise. Bad data is less easy to spot using independent line graphs for each brainwave frequency sometimes. The challenge for getting an entire meditation or sleep session’s spectrogram is that you need to record the raw EEG data at a high enough sample rate, fortunately the mind monitor app allows you to do just that. The exported file is zipped before being uploaded to your choice of internet based storage platform, which is good because unzipped, the file sizes can get pretty massive after a while, I’d advise stripping out fields you don’t need from the settings before recording raw EEG data.

In order to process and view this data, I personally found that audio spectrogram software works quite well as a starter. Here’s an example using Spek, which is a simple free tool:

The line that runs across, which appears almost halfway up the spectrogram is just noise from nearby AC line current, which is generally unavoidable unless you’re careful about keeping away from sockets and power lines etc, you can actually remove this using a notch filter in the mind-monitor app settings. Or just literally screen grab the image and cut it off :) example below from a different night:

One downside with some of these apps is that they’re generally tailored for audio signals which are at higher frequencies than what we’re looking. Some other programs such as Audacity allow you to modify the view in a lot of different ways so you can see more detail, you can see examples on my GitHub page here:

Beyond this basic way of viewing EEG data I do think it’s worth experimenting with using something like multitaper methods and morlet wavelets to really home in on details. One repo I found useful is here:

And here’s an example image using this with a few tailored settings over a night’s sleep data:

As you can see there’s better detail from the signal in here and also the axis is scaled more appropriately and it's clearer to see certain artifacts. It's easier to see more nuanced details when you zoom in and scroll across rather than all compressed into a small graph like this as well e.g. when you zoom in it's more obvious where jaw clenching occurs and possible markers for eye movement.

From this spectrogram you can easily see any problem areas in the data and with a bit of filtering, view periods of light and deep sleep. If you crosscheck with HR and gyro signals, both of which can also be obtained from the Muse data separately, you can have a better guess at REM stages too.

Another tool I check (also free) was EDFbrowser, you can convert wav files to EDF easily in here and then the tool has spectrogram window that has generally been built with EEG analysis in mind:

There’s probably much better ways to enhance the signal so I'm definitely not an expert and also I think the signal strength from the Muse band seems low compared to lab grade EEG devices.

Overall I think this is a great product for both meditation and sleep, it doesn’t really have any competition at the moment at this consumer level. I do think the software from Muse doesn’t provide as much as some of us would like in terms of exporting more low-level details but what we have available from mind-monitor and the ability to export data from there fills a big gap along with it's own additional interesting features.

1,142 views0 comments

Recent Posts

See All


bottom of page