It's nice that they sell a replacement band for gen 2 that's compatible with gen 1. This would help improve the signal on temporal lobes for the original Muse S.
I didn't know they do this - good spot! I would've probably just got the updated band in that case rather than the Gen 2 as it's the only real difference
@@parodoxis Yes, I bought gen 2 band and it works flawlessly on my first generation device. Please note that they still need to improve the sturdiness of the band.
Hi thanks for the video I have the Muse S first gen I was put off with the sleep tracking once when I spent about a half an hour awake in bed and checked the tracking and it said I had been in deep sleep some of the time. I still like the device for meditation though,. I have recently got an Mendi, this is a great device for focus, really felt the benefit straight away.
I have the Gen 1 but purchased a Gen 2 replacement band, which fits the Gen 1 device. Much larger sensors on the band. However, side sensors are still a problem and I have to pretty much soak the sides of my head to improve contact. I don’t see any reason upgrade to Gen 2, but will likely replace the band because sensors wear out after a few years.
Interesting video. Thanks. The sleep stages look really terrible in the app. According to that you’ve been awake almost all night. Even the worst wristband would be better than that. I would have thought the eeg would make it great. But it suffers so much from bad signals that it’s pretty much useless, or I’m a wrong? I’ve got a Dreem 2 (also eeg) and it does a lot better, but isn’t supported anymore, so I’m looking for an alternative, and thought I found it in the muse S, but you’ve shown in this video that it doesn’t come close.. or I’m I wrong ?
If you look at my latest video you'll see the results were better with the Gen 2, their algorithm does seem to work ok if you get sensor coverage and is definitely better than wrist based trackers. Signal quality does seem to vary with people, even with me but from my experience I would recommend a Gen 2 Muse S still despite most of my criticisms which are really around being able to see more of the data than just a hypnogram and basic HR/movement graphs. The Gen 1 Muse S is not good for consistancy and is now discontinued anyway I believe. I have a Dreem coming soon and I think there may be a way to work around the lack of ongoing support for it so I will have an update on that in Feb most likely. The Dreem band is much more complicated then the Muse, mainly to get the FDA approval for medical use so it's not actually a fair comparison and I wouldn't expect a Muse to give you as good results for sleep analysis.
@@majorinput4217 thanks a lot for your reply. To be honest, I do think it’s a fair comparison. Sensor wise they are comparable, and they cost both €399 in Europe. And they both claim to accurately sense sleep stages with EEG. The Dreem is great but it often classifies me as being in REM while I’m really just lying awake. I do not know if it’s the algorithm or a sensor isn’t okay. Sadly one cannot check the sensor data. Really a shame that Dreem is discontinued. I’m very interested in anything you’ll find!
@@voodooparadox2513 Hi yes I have, been using it for the past week or so but still going through that early phase where they do an initial assessment. so far so good but will be a couple more weeks before I really get into the details of it.
Hi and thanks so much for the valuable videos! On the Dreem: it shows as not currently being sold to general public, do you have any leads on this? Would you buy it second hand? Also would add that one major point in favour or Dreem over Muse for me is the fact that Muse (as well as Heartmath & Oura sensors) use Bluetooth and that is literally in this case frying your brain... :/
Hi Filipa, I've been planning to put a video out on the Dreem I just haven't go round to it yet but I did a fairly detailed writeup on it that covers your points: www.majorinput.co.uk/post/a-closer-look-at-the-dreem-eeg-headband I don't agree with the brain frying aspect of bluetooth on Muse vs Dreem, in fact the Dreem is probably worse near your head if anything because it's a much more powerful device and uses way more power - like an order of magnitude more because it can't rely on a phone or something else to do the heavy computing. Basically in my view the Dreem has a great smart alarm and the app reports much better sleep metrics but you can't get raw sleep data directly from the device and it is no longer supported for consumers so my guess is eventually this is going to be a bricked device. I would not recommend getting a Dreem, Muse has a better future for consumers and it's shortcomings can be improved through better software.
Does it have bone conducting speakers like Dreem 2 does through which it plays meditation sounds? Or do you need to use headphones for that? And can it be used like a smart alarm as Dreem 2 can? I don't really care about stats to be honest, but do want a device that could help me fall to sleep and wake me up during the best stage of my sleep. Can Muse S do that?
No bone conducting speakers but I guess you could use headphones as all audio features come from the phone rather than the device itself. There is no smart alarm in the Muse app and I'm really surprised that still isn't available because I don't see why it would be hard for them to put in. The Muse app does have some features that are supposed to aid falling asleep and staying in deep sleep but I've not really tested those as much. The Dreem band has a very good smart alarm and is the best I've used but right now I don't recommend anyone buying these because they are not supported for consumers anymore, I'll explain in more detail in another video.
It's something I've been looking into but with the current commercial apps it's not ideal - definitely not possible with the standard Muse app. It has a great set of sensors to be able to do that kind of thing but as I mentioned, without a supported SDK for people to make apps like that it's not so easy to do. And Muse's terms and process of approval for 3rd party apps seems convoluted based on the fact there's barely anything else besides mind monitor. Even streaming from Mind Monitor wouldn't work well for this, the best way currently is to use alternatives tools like MuseLSL so that you can get the data from the PPG sensor as well as EEG but so far I've had connection issues with those tools. If I can work that out then there's definitely a way forward but will involve connecting to a PC or at least something like a Raspberry PI to process data. github.com/alexandrebarachant/muse-lsl
It's a pity the narrator doesn't bother to speak clearly. I had to concentrate on deciphering his speech instead of understanding what he was talking about. The main problem, I suppose, is the usual one with electrical recording of nervous activity: the electrode impedance is high. I have done some more comparisons of the Muse S (Gen 2) sleep recording with video and heart rate monitoring and general motion recordings. I think the Muse S (Gen 2) has erratic performance as an EEG sleep monitor. I don't know whether the Muse S (Gen 2) really has much efficacy as an analyser of EEG for sleep recording. It is often enough quite wrong in its guesses of whether one is asleep, though it isn't pure guesswork: it does have some skill in detecting sleep/wakefulness. The "sleep staging" (normal/REM sleep) is erratic, perhaps devoid of skill. I have detected no deep sleep in my own records; perhaps I don't get deep sleep? To make the electrodes above the ears work, I have taken to shaving the area as well as thoroughly moistening it. A problem is that if one wants to record sleep every night, one has to think of skin care at the electrode contact areas. For really good contact, for example with gel electrodes, one needs to nearly scrub the skin with isopropyl alcohol swabs or some such; because that is rough on the skin, one can't do it every night, or even on successive nights. The Muse S (Gen 2) makes a fair compromise: accepting a relatively poor contact with high electrode impedance, while avoiding scrubbing the skin. There isn't local pressure on the skin for the above-ear electrodes. I guess that the forehead electrodes work better because they have more local pressure on them? Edit: well, well, well !!! Last night, for the first time, my Muse S (Gen 2) reported some "deep sleep" for me: 52 min 2 sec of it, broken into 5 or 6 segments with brief interruptions. Once I took off the headband to wet my skin, and, lo and behold, the heartbeat icon kept beating, and the midforehead electrode icon stayed full on green !! Hmm. Mostly, the heartbeat icon doesn't beat, even though the supporting red icon is on. I am more mystified as to how the device registers sleep: I guess that it's mainly by the head accelerometers, with scarce contribution from the heartbeat and EEG signals?
Apologies the audio isn't clear for you. I am making changes to my mic setup for the next videos I plan. Also for anyone else that has issues with the audio clarity or my accent, my website always includes an associated blog writeup that has the same detail for what is covered in my videos, there are also a few posts that I don't always convert into a video: www.majorinput.co.uk/blog In regards to the temporal electrodes - these are not great I agree. I think Muse will need to look at designing flexible comb style electrodes that can work through hair better for these. I've never had to do anything other than use a little bit of water for the front electrodes to work fine. If you want to reduce imedance further you could try conductive paste.
Thank you for your gracious response. To be frank, I think you should speak with care to articulate your words more distinctly. As for the conductive paste, thank you for the tip. It's a pain, but I will try it.
I guess they should maximise the contact area of the above-ear electrodes, much extend them. Perhaps they could double the padding of the band over them, to sharpen the pressure? The heartbeat thing is not good.
My observation is that the heartbeat record provides good information about sleep, though I doubt that many devices use it? Beyond the obvious persistent slowing, I observe three types of heartbeat-registered sleep. (2) rhytmic, perhaps 4 cycles per minute; (2) flat, little variation of rate; (3) erratic rate variation. I have little idea how those relate to the EEG stages, but I guess that (3) might signal REM sleep? It would be a feat to program this from a record, though often it's easy enough to see it by eye.
stumbled on this video after finding out there is a gen2 museS out so i wanted to know if i have to spend all this money again - happy to see i wont since i shave my hair 🤣🤣🤣 but i am using anyway most of the time my muse1 since it has a more accurate signal,muse2 was horrible coz of bad connections ,and both muse2 and S have inaccurate brainwaves with a distinct difference to muse1 ,in mind monitor graphs its about +20 for gamma , +15 for beta , +10 for theta and alpha while delta is same... i guess they did it on purpose to stop their app from giving wrong results coz of elevated delta which is seen by the app as sleep so they upped all other waves in comparison,and thats probably why they say that museS is not suited for research purposes... so you do all of this to detect sleep? there is a much easier way to do it by just using the mm graph...show me any graph and i tell you if you were sleeping or even dreaming as both of those states have a very specific pattern which stands out from all the rest... and then you either use alex' code to create a neurofeedback app since you seem to know abotu how to program ,or you can use neurovisual which has audio and/or visual neurofeedback already built into it where you can choose the brainwaves you want and the dB you want to trigger a sound of your choice while you stream data from mm to it,which works fine coz i used it for getting feedback on low gamma which i wanted to achieve.... www.mindresearch.xyz and here are the sleep and dream patterns... its not a matter of bad signal quality that the muse app gives wrong results but coz their algorithm doesnt work as it should,just as with the meditation app where when you go deeper it totally freaks out 🤣 imgur.com/jhPOsXz
Hi, Muse is not suitable for clinical use simply because the voltage amplifiers they use just don't have the right ammount of signal to noise ratio and bit depth. They've managed to create something that works good enough for consumers and basic hypnograms etc at a lower cost. I think REM is more complicated than just looking at discreet brainwaves, I actually think looking at the raw brainwaves for signals like k-complex and sleep spindles is better, which is why machine learning algorythms based on the time series data tends to work very well.
@@majorinput4217 if machine learning works well then interaxon's algorithm would work well,which they dont,or not good enough... yes,it is that simple as looking at a graph,i have analyzed and learned from literally thousands of graphs and people and i have found the sleep and dreams patterns which work 100% of the time without mistake,but dont take my word for it,go to sleep and let mm record and then look at the graph and you will see the same exact patterns i showed you in the picture,and when you have lucid dreams you will see delta become dominant.... same as meditative states,the high delte and /or theta the deeper the experience,specially when they are on 2nd place right below alpha,and low gamma showing stillness of mind,every single time,the more you can still your mind the lower gamma will be.... i dont know who gave you the info about muse not being suitable for research but obviously thats not true coz there are tons of papers out there proving the muse1 is good enough for research,to a point where this neuroscientist who is pretty popular and even worked for nasa with the muse has compared it to a super expensive research grade EEG at his university which costs hundreds of thousands and said they both basically gave the same results > ruclips.net/video/AaJpk2w4JBc/видео.html
It's nice that they sell a replacement band for gen 2 that's compatible with gen 1. This would help improve the signal on temporal lobes for the original Muse S.
I didn't know they do this - good spot! I would've probably just got the updated band in that case rather than the Gen 2 as it's the only real difference
Is it really??
@@parodoxis Yes, I bought gen 2 band and it works flawlessly on my first generation device. Please note that they still need to improve the sturdiness of the band.
Really interesting and helpful video, thank you! Can’t wait to see the VR/gaming industries develop in this area, as you say.
Hi thanks for the video I have the Muse S first gen I was put off with the sleep tracking once when I spent about a half an hour awake in bed and checked the tracking and it said I had been in deep sleep some of the time.
I still like the device for meditation though,.
I have recently got an Mendi, this is a great device for focus, really felt the benefit straight away.
The Mendi looks very interesting, been keeping an eye on that one and fNIRS in general
I have the Gen 1 but purchased a Gen 2 replacement band, which fits the Gen 1 device. Much larger sensors on the band. However, side sensors are still a problem and I have to pretty much soak the sides of my head to improve contact. I don’t see any reason upgrade to Gen 2, but will likely replace the band because sensors wear out after a few years.
Hi is the heart rate sensor more accurate with the protuding pod?
I think the HR sensor was fine before for me, this shape probaby helps if you had issues previously
Does the S 1st generation have better brain connection and signal than the regular Muse 2 band?
yes, no contest
Can't get the thing (s2) to consistently connect to my phone, going to have to return it.
Interesting video. Thanks. The sleep stages look really terrible in the app. According to that you’ve been awake almost all night. Even the worst wristband would be better than that. I would have thought the eeg would make it great. But it suffers so much from bad signals that it’s pretty much useless, or I’m a wrong? I’ve got a Dreem 2 (also eeg) and it does a lot better, but isn’t supported anymore, so I’m looking for an alternative, and thought I found it in the muse S, but you’ve shown in this video that it doesn’t come close.. or I’m I wrong ?
If you look at my latest video you'll see the results were better with the Gen 2, their algorithm does seem to work ok if you get sensor coverage and is definitely better than wrist based trackers.
Signal quality does seem to vary with people, even with me but from my experience I would recommend a Gen 2 Muse S still despite most of my criticisms which are really around being able to see more of the data than just a hypnogram and basic HR/movement graphs. The Gen 1 Muse S is not good for consistancy and is now discontinued anyway I believe.
I have a Dreem coming soon and I think there may be a way to work around the lack of ongoing support for it so I will have an update on that in Feb most likely. The Dreem band is much more complicated then the Muse, mainly to get the FDA approval for medical use so it's not actually a fair comparison and I wouldn't expect a Muse to give you as good results for sleep analysis.
@@majorinput4217 thanks a lot for your reply. To be honest, I do think it’s a fair comparison. Sensor wise they are comparable, and they cost both €399 in Europe. And they both claim to accurately sense sleep stages with EEG. The Dreem is great but it often classifies me as being in REM while I’m really just lying awake. I do not know if it’s the algorithm or a sensor isn’t okay. Sadly one cannot check the sensor data. Really a shame that Dreem is discontinued. I’m very interested in anything you’ll find!
Have you received the Dreem already? Exited to see a comparison.
@@voodooparadox2513 Hi yes I have, been using it for the past week or so but still going through that early phase where they do an initial assessment. so far so good but will be a couple more weeks before I really get into the details of it.
@@majorinput4217 great! Excited to hear what you think!
Hi and thanks so much for the valuable videos! On the Dreem: it shows as not currently being sold to general public, do you have any leads on this? Would you buy it second hand? Also would add that one major point in favour or Dreem over Muse for me is the fact that Muse (as well as Heartmath & Oura sensors) use Bluetooth and that is literally in this case frying your brain... :/
Hi Filipa, I've been planning to put a video out on the Dreem I just haven't go round to it yet but I did a fairly detailed writeup on it that covers your points: www.majorinput.co.uk/post/a-closer-look-at-the-dreem-eeg-headband
I don't agree with the brain frying aspect of bluetooth on Muse vs Dreem, in fact the Dreem is probably worse near your head if anything because it's a much more powerful device and uses way more power - like an order of magnitude more because it can't rely on a phone or something else to do the heavy computing. Basically in my view the Dreem has a great smart alarm and the app reports much better sleep metrics but you can't get raw sleep data directly from the device and it is no longer supported for consumers so my guess is eventually this is going to be a bricked device. I would not recommend getting a Dreem, Muse has a better future for consumers and it's shortcomings can be improved through better software.
It's not classic Bluetooth, it's Bluetooth Low Energy
Thanks!
Does it have bone conducting speakers like Dreem 2 does through which it plays meditation sounds? Or do you need to use headphones for that?
And can it be used like a smart alarm as Dreem 2 can?
I don't really care about stats to be honest, but do want a device that could help me fall to sleep and wake me up during the best stage of my sleep. Can Muse S do that?
No bone conducting speakers but I guess you could use headphones as all audio features come from the phone rather than the device itself. There is no smart alarm in the Muse app and I'm really surprised that still isn't available because I don't see why it would be hard for them to put in. The Muse app does have some features that are supposed to aid falling asleep and staying in deep sleep but I've not really tested those as much. The Dreem band has a very good smart alarm and is the best I've used but right now I don't recommend anyone buying these because they are not supported for consumers anymore, I'll explain in more detail in another video.
@@majorinput4217 thanks for your reply. Sounds like it is not the device I am looking for
Hello, I want to buy one for my nephew, how do I do it?
How do you know what color for which wave
Can you use it for lucid dreaming or hypnogogia induction?
It's something I've been looking into but with the current commercial apps it's not ideal - definitely not possible with the standard Muse app. It has a great set of sensors to be able to do that kind of thing but as I mentioned, without a supported SDK for people to make apps like that it's not so easy to do. And Muse's terms and process of approval for 3rd party apps seems convoluted based on the fact there's barely anything else besides mind monitor. Even streaming from Mind Monitor wouldn't work well for this, the best way currently is to use alternatives tools like MuseLSL so that you can get the data from the PPG sensor as well as EEG but so far I've had connection issues with those tools. If I can work that out then there's definitely a way forward but will involve connecting to a PC or at least something like a Raspberry PI to process data.
github.com/alexandrebarachant/muse-lsl
It's a pity the narrator doesn't bother to speak clearly. I had to concentrate on deciphering his speech instead of understanding what he was talking about. The main problem, I suppose, is the usual one with electrical recording of nervous activity: the electrode impedance is high. I have done some more comparisons of the Muse S (Gen 2) sleep recording with video and heart rate monitoring and general motion recordings. I think the Muse S (Gen 2) has erratic performance as an EEG sleep monitor. I don't know whether the Muse S (Gen 2) really has much efficacy as an analyser of EEG for sleep recording. It is often enough quite wrong in its guesses of whether one is asleep, though it isn't pure guesswork: it does have some skill in detecting sleep/wakefulness. The "sleep staging" (normal/REM sleep) is erratic, perhaps devoid of skill. I have detected no deep sleep in my own records; perhaps I don't get deep sleep?
To make the electrodes above the ears work, I have taken to shaving the area as well as thoroughly moistening it. A problem is that if one wants to record sleep every night, one has to think of skin care at the electrode contact areas. For really good contact, for example with gel electrodes, one needs to nearly scrub the skin with isopropyl alcohol swabs or some such; because that is rough on the skin, one can't do it every night, or even on successive nights. The Muse S (Gen 2) makes a fair compromise: accepting a relatively poor contact with high electrode impedance, while avoiding scrubbing the skin. There isn't local pressure on the skin for the above-ear electrodes. I guess that the forehead electrodes work better because they have more local pressure on them? Edit: well, well, well !!! Last night, for the first time, my Muse S (Gen 2) reported some "deep sleep" for me: 52 min 2 sec of it, broken into 5 or 6 segments with brief interruptions. Once I took off the headband to wet my skin, and, lo and behold, the heartbeat icon kept beating, and the midforehead electrode icon stayed full on green !! Hmm. Mostly, the heartbeat icon doesn't beat, even though the supporting red icon is on. I am more mystified as to how the device registers sleep: I guess that it's mainly by the head accelerometers, with scarce contribution from the heartbeat and EEG signals?
Apologies the audio isn't clear for you. I am making changes to my mic setup for the next videos I plan. Also for anyone else that has issues with the audio clarity or my accent, my website always includes an associated blog writeup that has the same detail for what is covered in my videos, there are also a few posts that I don't always convert into a video: www.majorinput.co.uk/blog
In regards to the temporal electrodes - these are not great I agree. I think Muse will need to look at designing flexible comb style electrodes that can work through hair better for these. I've never had to do anything other than use a little bit of water for the front electrodes to work fine. If you want to reduce imedance further you could try conductive paste.
Thank you for your gracious response. To be frank, I think you should speak with care to articulate your words more distinctly. As for the conductive paste, thank you for the tip. It's a pain, but I will try it.
I guess they should maximise the contact area of the above-ear electrodes, much extend them. Perhaps they could double the padding of the band over them, to sharpen the pressure? The heartbeat thing is not good.
My observation is that the heartbeat record provides good information about sleep, though I doubt that many devices use it? Beyond the obvious persistent slowing, I observe three types of heartbeat-registered sleep. (2) rhytmic, perhaps 4 cycles per minute; (2) flat, little variation of rate; (3) erratic rate variation. I have little idea how those relate to the EEG stages, but I guess that (3) might signal REM sleep? It would be a feat to program this from a record, though often it's easy enough to see it by eye.
Terrible sensors on Gen 1
You can order a Gen 2 replacement band, which fits the Gen 1 pod. Problem solved.
stumbled on this video after finding out there is a gen2 museS out so i wanted to know if i have to spend all this money again - happy to see i wont since i shave my hair 🤣🤣🤣
but i am using anyway most of the time my muse1 since it has a more accurate signal,muse2 was horrible coz of bad connections ,and both muse2 and S have inaccurate brainwaves with a distinct difference to muse1 ,in mind monitor graphs its about +20 for gamma , +15 for beta , +10 for theta and alpha while delta is same... i guess they did it on purpose to stop their app from giving wrong results coz of elevated delta which is seen by the app as sleep so they upped all other waves in comparison,and thats probably why they say that museS is not suited for research purposes...
so you do all of this to detect sleep? there is a much easier way to do it by just using the mm graph...show me any graph and i tell you if you were sleeping or even dreaming as both of those states have a very specific pattern which stands out from all the rest... and then you either use alex' code to create a neurofeedback app since you seem to know abotu how to program ,or you can use neurovisual which has audio and/or visual neurofeedback already built into it where you can choose the brainwaves you want and the dB you want to trigger a sound of your choice while you stream data from mm to it,which works fine coz i used it for getting feedback on low gamma which i wanted to achieve.... www.mindresearch.xyz
and here are the sleep and dream patterns... its not a matter of bad signal quality that the muse app gives wrong results but coz their algorithm doesnt work as it should,just as with the meditation app where when you go deeper it totally freaks out 🤣
imgur.com/jhPOsXz
Hi, Muse is not suitable for clinical use simply because the voltage amplifiers they use just don't have the right ammount of signal to noise ratio and bit depth. They've managed to create something that works good enough for consumers and basic hypnograms etc at a lower cost. I think REM is more complicated than just looking at discreet brainwaves, I actually think looking at the raw brainwaves for signals like k-complex and sleep spindles is better, which is why machine learning algorythms based on the time series data tends to work very well.
@@majorinput4217 if machine learning works well then interaxon's algorithm would work well,which they dont,or not good enough...
yes,it is that simple as looking at a graph,i have analyzed and learned from literally thousands of graphs and people and i have found the sleep and dreams patterns which work 100% of the time without mistake,but dont take my word for it,go to sleep and let mm record and then look at the graph and you will see the same exact patterns i showed you in the picture,and when you have lucid dreams you will see delta become dominant....
same as meditative states,the high delte and /or theta the deeper the experience,specially when they are on 2nd place right below alpha,and low gamma showing stillness of mind,every single time,the more you can still your mind the lower gamma will be....
i dont know who gave you the info about muse not being suitable for research but obviously thats not true coz there are tons of papers out there proving the muse1 is good enough for research,to a point where this neuroscientist who is pretty popular and even worked for nasa with the muse has compared it to a super expensive research grade EEG at his university which costs hundreds of thousands and said they both basically gave the same results > ruclips.net/video/AaJpk2w4JBc/видео.html