I have hooked up the Remedy between my Logitech Squeezebox and my Audio Research DAC 8. The only true jitter measurements I could find were for it's slightly newer cousin, the DSPre. It seems that the DSPre is wildly sensitive to jitter, at least the measurements are, so I'm just going to assume the DAC 8 is the same or worse.
Most of my listening these days is to Internet radio, including Toronto Jazz FM 91, which in addition to having great programming also streams at 24/96. The other station was KDFC 90.1, Bay Area Classical.
In the middle of this my power regulator has started to hum, so I can't do the remedy justice until I move it to a quieter location.
Still, here's what I think so far.
The remedy works much more noticeably with low-resolution stations, but it IS better. The sense of space inside the sound stage and the treble decay. Sometimes it feels worse though. During mass string crescendos the sound gets too complicated. With Jazz 91 the improvements seem much less pronounced, but still there.
However this is really hard to gauge with radio. I'll give this a better listen soon, when my biggest noise sources have been fixed.
Update August 28, 2016
So I had thought my impressions might greatly change, but they did not. I think this is a good tool but only for more source sensitive DAC's like I had (just sold it), the ARC DAC 8. That DAC played brilliantly when driven by an Ayre CD player, but when I brought it home it was much fussier.
With high resolution music (96k and above) I could honestly not tell if it was working or not, which I guess is a good thing. So, overall I would recommend this to clean up the sound of a mid-Fi CD player or Internet radio or inexpensive streamer like my Squeezebox Touch.
I have however switched over to a Mytek Brooklyn DAC which sounds as good as the Wyred4Sound + ARC DAC 8 without the remedy in place, regardless of how it was driven.
I'm now driving the Brooklyn with a 2TB Linux streamer I built myself ($650) and it's very happy to play PCM, MQA and DSD from it.
Saturday, July 16, 2016
Monday, July 11, 2016
Digital Audio - Upsampling and Oversampling Explained
Many types of digital sources, accessories and Digital to Analog Converters (DACs) provide some sort of sample data magic called oversampling or upsampling. Put simply it means you end up with more digital data than you started with.
There are some benefits, but none of these methods truly gets you closer to the original music. They are all just ways of trying to make the experience more pleasant. Think of it as looking out your window with a screen. You may take a picture and find that you can see the screen itself in the image, or you can do some editing with Gimp or PhotoShop and remove it. The new image can't possibly contain more true to life data than you started with, but the picture should be much more pleasant to look at.
Many audiophiles have been led to believe that this kind of digital math can do things like you might see on the TV shows CSI or NCIS. Somehow four pixels on a grainy satellite image can be processed over and over again until the criminal's face is clearly visible. It's just not true.
Looking at it another way, the frequency response of up and oversampling does not change. A 44.1 kHz file is not going to have 30kHz created after 4x upsampling. The frequency range and content density is unchanged. What may happen is that digital filtering becomes smoother and easier on the ears, or that jitter is improved somewhat by the use of higher data rates.
at a sample rate of 44.1 kHz these two samples represent:
So, instead of 2 samples, we have 8, but with a new sample rate. Lets redo the math:
It's so simple you don't think it should work. Take a sample, and repeat it several times. It's that simple. It does not attempt to provide any more data but may shift some noise far above the Nyquist frequency. No math is involved, just counting. With 4x oversampling the DAC our orignal two samples become:
Things get even more muddled when upsampling is used for ASRC, Asynchronous Sample Rate Conversion, but also more beneficial, as it's one of the best ways to reduce jitter. More on that in a future post.
Technically and mathematically more challenging, there are two general approaches. To take the best advantage of this it's better if the bit depth increases beyond the original. So if the original was 16 bits, 24 or 32 bits will provide better resolution. However remember that this doesn't really make it more true to life. It just makes some things easier to do and helps us keep more of our results. There are some VERY nice 32 bit DAC chips out there though, so taking full advantage of them may also get us much closer to true 24 bit resolution. That's a topic for someone else.
Again, consider our original two samples, 24 and 28. The rate of change is 4/sample. 4/4 = 1. Now the DAC gets:
Technically you only need 2 samples for a spline, but the result is the same as linear interpolation, so we'll ignore that case. With spline math we take a number of samples, usually under 20, to draw a much softer curve. Wadia was the first company I know of who introduced this concept. In this case it really helps to have more bits, as the extra bits help with more fine grained results. As you might imagine, the math and CPU power required is greatest for this example.
If this was floating point math our working data set would be:
There are some benefits, but none of these methods truly gets you closer to the original music. They are all just ways of trying to make the experience more pleasant. Think of it as looking out your window with a screen. You may take a picture and find that you can see the screen itself in the image, or you can do some editing with Gimp or PhotoShop and remove it. The new image can't possibly contain more true to life data than you started with, but the picture should be much more pleasant to look at.
Many audiophiles have been led to believe that this kind of digital math can do things like you might see on the TV shows CSI or NCIS. Somehow four pixels on a grainy satellite image can be processed over and over again until the criminal's face is clearly visible. It's just not true.
Looking at it another way, the frequency response of up and oversampling does not change. A 44.1 kHz file is not going to have 30kHz created after 4x upsampling. The frequency range and content density is unchanged. What may happen is that digital filtering becomes smoother and easier on the ears, or that jitter is improved somewhat by the use of higher data rates.
Differences Explained
Let's take original data. Since digital music is always integer, I'll imagine two consecutive samples with convenient values of 24 and 28. Now lets see what happens at 4x up or oversampling. If the original data was 44kHz/16 bits the DAC will now see a sample rate of 176.4 kHz but the bit depth may or may not. So, just to be thorough, here is our original data:- 24
- 28
at a sample rate of 44.1 kHz these two samples represent:
- 2 samples / 44,100 = 45 microseconds of music.
So, instead of 2 samples, we have 8, but with a new sample rate. Lets redo the math:
- 8 samples / 176.4kHz = 45 microseconds of music.
Oversampling
This is the oldest trick in the book. Almost immediately after CD players became commercially available oversampling became a buzz-word. I am no longer sure, but this may have only worked with so-called Delta-Sigma or 1-bit DAC's.It's so simple you don't think it should work. Take a sample, and repeat it several times. It's that simple. It does not attempt to provide any more data but may shift some noise far above the Nyquist frequency. No math is involved, just counting. With 4x oversampling the DAC our orignal two samples become:
- 24
- 24
- 24
- 24
- 28
- 28
- 28
- 28
Upsampling
Bit Perfection
One of the objections to upsampling, is that the signal is no longer bit-perfect. The DAC no longer gets the original facts, but the original facts, plus a lot more. That "lots more" is pure mathematical conjecture. However, there are some real benefits to be had.Things get even more muddled when upsampling is used for ASRC, Asynchronous Sample Rate Conversion, but also more beneficial, as it's one of the best ways to reduce jitter. More on that in a future post.
Technically and mathematically more challenging, there are two general approaches. To take the best advantage of this it's better if the bit depth increases beyond the original. So if the original was 16 bits, 24 or 32 bits will provide better resolution. However remember that this doesn't really make it more true to life. It just makes some things easier to do and helps us keep more of our results. There are some VERY nice 32 bit DAC chips out there though, so taking full advantage of them may also get us much closer to true 24 bit resolution. That's a topic for someone else.
Linear Interpolation
Imagine two points on a chart. Draw a straight line between them. That's simple interpolation. It's no more complicated than simple algebra. Calculate the rise, divide it by the number of intervening samples, and add that much for each "new" sample. For linear interpolation, the sample rate converter needs to know two samples at a time in order to figure out the rate at which the intermediate samples should change.Again, consider our original two samples, 24 and 28. The rate of change is 4/sample. 4/4 = 1. Now the DAC gets:
- 24 +1 =
- 25 +1 =
- 26 +1 =
- 27 +1 =
- 28
Spline
A much more advanced way to create more samples is by using what are called splines. Remember the "French Curve" tools you may have used in drawing school?
Technically you only need 2 samples for a spline, but the result is the same as linear interpolation, so we'll ignore that case. With spline math we take a number of samples, usually under 20, to draw a much softer curve. Wadia was the first company I know of who introduced this concept. In this case it really helps to have more bits, as the extra bits help with more fine grained results. As you might imagine, the math and CPU power required is greatest for this example.
If this was floating point math our working data set would be:
- (nine samples before)
- 24.000
- 25.185
- 26.355
- 27.888
- 28.000
- (nine samples after)
Are Splines Really Better?
Splines are very cool, but it may be argued, convincingly, that we are not doing much more than you could achieve with a capacitor and resistor with the proper time constants. In other words, it's a lot of math and hardware for what could be done with $2 or less in parts. The real potential benefit of this advanced though is in custom algorithms. You can be as creative as you want to in your algorithms.What About Sound Quality?
Personally I have come to believe that the analog output stages matter much more than interpolating algorithms and sample rates or bit depth but the devil is in the implementation details. As always, buy what you like, and what is most pleasant to your ears. Don't buy algorithms or chips. Buy results, and spend no money that isn't pleasing to you.Wyred4Sound Remedy - Snake Oil or a True Panacea?
I've just ordered a Wyred4Sound Remedy. I had been drinking and needed a pick-me-up and based on Digital Audio Review's positive impressions I ordered it.
What I have just realized however is that the Remedy is not the product I thought it was in a couple of ways.
Mind you, it's clear that for some kinds of low-grade digital audio sources the Remedy is probably a very good solution. I'm thinking of Sonos, Apple TV and Chromecast specifically. However, it is a terribly over-hyped product which smells of snake oil.
Using an ASRC is a very good way (and a little lazy by itself) to ensure minimal jitter with possibly very jittery sources, such as Internet radio. The reason I'm kind of on the fence about this is that an ASRC is no longer bit perfect, but time-perfect. To ensure that every x picoseconds a new sample is processed, regardless of how the input signal may vary in the short and long term an ASRC resorts to a mathematical brute-force method, the details of which are beyond the scope of this posting. Suffice it to say you can kiss bit-perfection goodbye, and not just for the interpolated samples either.
One major annoyance, that relegates the Remedy to mid-fi sources is that the input signal is ALWAYS recreated. What's worse is that if you have music with a higher sample rate, such as 24/192kHz, the Remedy will actually DOWN-sample it to 24/96kHz. This, plus having no input switching makes his solution seem kind of dopey.
For about three times more a more robust option is the Mytek Stereo 192 SRC. It gives you much better control over what you want to do with the signal, up, down or no change with equivalent or better jitter reduction, as well as being able to convert up to 24/192kHz
Of course, this is all spec-manship. Listening is the true arbiter of what you should buy.
Having examined one, it seems W4S has used one of the best commonly available Crystek oscillators, the CCHD-957 series, which does in fact have very good phase noise characteristics, among those affordable to mere mortals. I'm still confused though. Based on W4S's own measurements this does not appear to be a femto-clock grade solution but a pico clock. What's the difference? About 1,000 times worse performance. It is possible that the internal clock device inside the case is a femto-clock class part, but that the other circuitry used can't take full advantage of it, or that it can only do so much in one pass. It would be very interesting to see measured comparisons using a standard Mac Mini or Apple TV to see how it measures to Mytek, M2Tech or Auralic with and without.
Consider this. The Auralic Vega with a true femto clock (and 10x more expensive) has jitter around 80 femto seconds. The image on W4S's own Remedy page shows jitter around 87 pico seconds. That's about 1,000 times worse performance. Of course, many would argue that you can't hear 80 pico seconds of jitter, but the point is the marketing hype. I don't like being lied to or misled.
Another similar device with a price point kind of in between is the M2Tech HiFace Evo 2. It is intended as a USB to SPDIF interface, but it will also take a coaxial SPDIF as an iput and allow you to select sample rate conversion. Price is around $700 USD.
It's a little odd as many DAC's make upsampling a key feature. They charge more for it and often tout their proprietary algorithms as being better one way or another.
In the end though I'll have to listen to it to evaluate the Remedy as having any sort of meaningful benefit. More on that in the next several weeks. One of my sources however will be Toronto Jazz 91 which streams at 24/96kHz already, this will let me do direct comparison to see if the jitter reduction is worthwhile, in addition to a small selection of 24/96kHz FLAC recordings I have.
Many listeners are easily swayed by "different." Even John Coltrane suffered this, always thinking his next performance was better. If jitter or SRC at these levels is audible it's quite possible many will be swayed by a different sound, but not necessarily a better sound. I can imagine many will get worse jitter than they started with, and then proclaim how audible and beneficent the differences are!
What I have just realized however is that the Remedy is not the product I thought it was in a couple of ways.
Mind you, it's clear that for some kinds of low-grade digital audio sources the Remedy is probably a very good solution. I'm thinking of Sonos, Apple TV and Chromecast specifically. However, it is a terribly over-hyped product which smells of snake oil.
What Kind of Product is the Remedy?
Of course, marketing people, being devoid of souls at birth, are free to call a tomato a vacuum cleaner and there's rarely any legal consequences. In my world however the Remedy should properly be called a sample rate converter (SRC) or Asynchronous Sample Rate Converter, a feature built into many of the top DAC chips today. SRCs always includes re-clocking, so calling it an SRC with reclocker is redundant.Using an ASRC is a very good way (and a little lazy by itself) to ensure minimal jitter with possibly very jittery sources, such as Internet radio. The reason I'm kind of on the fence about this is that an ASRC is no longer bit perfect, but time-perfect. To ensure that every x picoseconds a new sample is processed, regardless of how the input signal may vary in the short and long term an ASRC resorts to a mathematical brute-force method, the details of which are beyond the scope of this posting. Suffice it to say you can kiss bit-perfection goodbye, and not just for the interpolated samples either.
One major annoyance, that relegates the Remedy to mid-fi sources is that the input signal is ALWAYS recreated. What's worse is that if you have music with a higher sample rate, such as 24/192kHz, the Remedy will actually DOWN-sample it to 24/96kHz. This, plus having no input switching makes his solution seem kind of dopey.
For about three times more a more robust option is the Mytek Stereo 192 SRC. It gives you much better control over what you want to do with the signal, up, down or no change with equivalent or better jitter reduction, as well as being able to convert up to 24/192kHz
Of course, this is all spec-manship. Listening is the true arbiter of what you should buy.
Is it really a femto-clock?
Consider this. The Auralic Vega with a true femto clock (and 10x more expensive) has jitter around 80 femto seconds. The image on W4S's own Remedy page shows jitter around 87 pico seconds. That's about 1,000 times worse performance. Of course, many would argue that you can't hear 80 pico seconds of jitter, but the point is the marketing hype. I don't like being lied to or misled.
Another similar device with a price point kind of in between is the M2Tech HiFace Evo 2. It is intended as a USB to SPDIF interface, but it will also take a coaxial SPDIF as an iput and allow you to select sample rate conversion. Price is around $700 USD.
Why does this matter?
My point to all of this is that the Remedy is doing more than just jitter reduction, and I would really have liked to know this before I ordered. Hiding major behavior is not a sign of a trustworthy vendor. Remedy is playing with the bits and I should have known that first. For more on why this is different, see my post on Upsampling and Oversampling.It's a little odd as many DAC's make upsampling a key feature. They charge more for it and often tout their proprietary algorithms as being better one way or another.
In the end though I'll have to listen to it to evaluate the Remedy as having any sort of meaningful benefit. More on that in the next several weeks. One of my sources however will be Toronto Jazz 91 which streams at 24/96kHz already, this will let me do direct comparison to see if the jitter reduction is worthwhile, in addition to a small selection of 24/96kHz FLAC recordings I have.
Many listeners are easily swayed by "different." Even John Coltrane suffered this, always thinking his next performance was better. If jitter or SRC at these levels is audible it's quite possible many will be swayed by a different sound, but not necessarily a better sound. I can imagine many will get worse jitter than they started with, and then proclaim how audible and beneficent the differences are!
Friday, July 8, 2016
So your CPAP treatment isn't working?
I've gone through seven or eight sleep studies. I've lost count of exactly how many, but the last three at the same sleep lab, without my symptoms being fully improved. I would estimate that at the beginning they were about 20-30% better, but years later seemed to fail completely.
What I eventually discovered is that exercise actually made my sleep quality much worse. If this sounds like you, it might, then it is possible we share an odd, and previously unreported condition. The only other two conditions which I've read that can cause this is Cushing's Syndrome and Chronic Fatigue. However, sleep apnea that is not fully treated can mimic both of these, including elevated cortisol levels. So if you have noticed, or are able to determine if exercise is bad for you, then we may share a condition, which is treatable!
The problem in my case only was that I have two modes of sleeping. The lazy day, fully rested mode, and the "I've worked out" mode.
I didn't notice the problem at the time, but looking back it seems that since I slept poorly after exercising I learned not to exercise, even a little. I stopped taking the stairs and always chose an elvator, had groceries and food delivered, anything I could do to avoid physical labor. That is, my symptoms conditioned me not to exercise, so I would go get a study done after a couple of days of being fully rested, and as a result my full symptoms did not occur. I only know this after-the-fact because of the use of a Jawbone UP (lasted 3 months, but did the job) which allowed me to discover exactly when I had good and bad days of sleep.
After using the Jawbone for several months looking for clues to my problem I was able to completely correlate exercise with poor sleep the next several nights. This is the exact opposite of what should happen for most people but for me was the wall I could not overcome.
Another thing that seems to trigger this effect is calcium channel blocking medications, or medications which are related. These can be prescribed for a range of issues from depression to high blood pressure. The good news is that if this is you then your sleep studies should show up without having to go through the exercise portion of this experiment.
Further, the post-exercise crashes would last for days, sometimes even a week.
If you have a Jawbone, FitBit or ResMed S+ you may see your deep sleep very disturbed, or lots of wakings, despite the CPAP measuring otherwise low AHI numbers and tolerable leaks.
In the picture on the left I share an image from the Jawbone UP application showing my very worst night of sleep. See how many different periods I have of deep and light sleep? There should be about 8 of them. Instead I have about 17. Also notice the time I slept. Over 11 hours! That's because the sleep quality was so poor.
You should have big chunks, around 6-9 of deep sleep/light sleep cycles with no awakenings.
If you want to investigate on your own first, try to correlate it over a couple of weekends. One weekend take it completely easy. Do no physical labor for two days, see how well you sleep Sunday night. If weekends are your housework days, hire a cleaning service this once. If you are already in a disturbed sleep cycle though you may need to do this for several days until your symptoms clear. Only after you are in a "good" cycle should you attempt to disturb your sleep with exercise.
The next weekend, attempt to exercise normally both days. I don't mean do your normal routine, I mean work out 30-45 minutes each day, see how much of a difference this makes to your sleep Sunday night. If you find that your first Monday was great, and your second Monday terrible, you fit this profile. However, there are many unanswered questions that remain, and I've noticed in addition to exercise a secondary, cyclical pattern which I have yet to pin down.
Notice that the ResMed S+ was less sensitive for me than the Jawbone for this particular issue.
It really helps to hire house keepers and have friends to support you. Let them know what you are up to and that you'll need extra help while you try this. Take time off from work if needed. If you are like me, you'll crash for 3-5 days and be completely unproductive after the exercise portions.
It's also important to stress that immediately after exercising I felt great. My lungs felt clear, I felt light on my feet, it was such a great thing. It was only after I slept and failed to recover that symptoms appear.
Right now as far as I know, there is no guidance on exercise before a sleep study/titration, however it's generally recommended not to exercise in the late afternoon evening as it can wake you up and prevent you from sleeping fully.
The day before the sleep study:
The day of the sleep study:
In my case, it was like magic, symptoms that had never shown up before became clearly and consistently visible during the lab work. The CPAP machine that was fine for the first 6 studies was completely inadequate now.
By challenging your body, and your CPAP machine you'll have the best study possible.
If you are a sleep specialist, or researcher, and would like detailed information I would be willing to make that available privately. Please leave me a comment with information about your publicly visible contact information and I'll reach out to you.
What I eventually discovered is that exercise actually made my sleep quality much worse. If this sounds like you, it might, then it is possible we share an odd, and previously unreported condition. The only other two conditions which I've read that can cause this is Cushing's Syndrome and Chronic Fatigue. However, sleep apnea that is not fully treated can mimic both of these, including elevated cortisol levels. So if you have noticed, or are able to determine if exercise is bad for you, then we may share a condition, which is treatable!
The problem in my case only was that I have two modes of sleeping. The lazy day, fully rested mode, and the "I've worked out" mode.
Why didn't the previous sleep studies help?
The previous sleep studies did find a partial solution.I didn't notice the problem at the time, but looking back it seems that since I slept poorly after exercising I learned not to exercise, even a little. I stopped taking the stairs and always chose an elvator, had groceries and food delivered, anything I could do to avoid physical labor. That is, my symptoms conditioned me not to exercise, so I would go get a study done after a couple of days of being fully rested, and as a result my full symptoms did not occur. I only know this after-the-fact because of the use of a Jawbone UP (lasted 3 months, but did the job) which allowed me to discover exactly when I had good and bad days of sleep.
After using the Jawbone for several months looking for clues to my problem I was able to completely correlate exercise with poor sleep the next several nights. This is the exact opposite of what should happen for most people but for me was the wall I could not overcome.
Confounds
Moderate exercise (5,000 steps / day) would not cause this problem to occur immediately. It would sneak up on me. I needed it to be vigorous, or weight bearing to cause the symptoms to show up faster.Another thing that seems to trigger this effect is calcium channel blocking medications, or medications which are related. These can be prescribed for a range of issues from depression to high blood pressure. The good news is that if this is you then your sleep studies should show up without having to go through the exercise portion of this experiment.
Further, the post-exercise crashes would last for days, sometimes even a week.
How can I tell if this fits me?
In the picture on the left I share an image from the Jawbone UP application showing my very worst night of sleep. See how many different periods I have of deep and light sleep? There should be about 8 of them. Instead I have about 17. Also notice the time I slept. Over 11 hours! That's because the sleep quality was so poor.
You should have big chunks, around 6-9 of deep sleep/light sleep cycles with no awakenings.
Be Your Own Guinea Pig
If you want to investigate on your own first, try to correlate it over a couple of weekends. One weekend take it completely easy. Do no physical labor for two days, see how well you sleep Sunday night. If weekends are your housework days, hire a cleaning service this once. If you are already in a disturbed sleep cycle though you may need to do this for several days until your symptoms clear. Only after you are in a "good" cycle should you attempt to disturb your sleep with exercise.
The next weekend, attempt to exercise normally both days. I don't mean do your normal routine, I mean work out 30-45 minutes each day, see how much of a difference this makes to your sleep Sunday night. If you find that your first Monday was great, and your second Monday terrible, you fit this profile. However, there are many unanswered questions that remain, and I've noticed in addition to exercise a secondary, cyclical pattern which I have yet to pin down.
Notice that the ResMed S+ was less sensitive for me than the Jawbone for this particular issue.
How can I get better?
Schedule another sleep study, but this time make sure you exercise normally the day of the study, if not two days before the study. In my case this was really a challenge, since my symptoms had been getting worse, my energy levels and moods had been suffering, causing me to exercise less and less.It really helps to hire house keepers and have friends to support you. Let them know what you are up to and that you'll need extra help while you try this. Take time off from work if needed. If you are like me, you'll crash for 3-5 days and be completely unproductive after the exercise portions.
It's also important to stress that immediately after exercising I felt great. My lungs felt clear, I felt light on my feet, it was such a great thing. It was only after I slept and failed to recover that symptoms appear.
Tips:
Remember that exercise IS recommended as part of almost every sleep improvement regimen, so you aren't breaking any rules by following the advice here. In all cases follow the advice of your doctors and personal trainers, but the guidelines here should be in compliance with them.Right now as far as I know, there is no guidance on exercise before a sleep study/titration, however it's generally recommended not to exercise in the late afternoon evening as it can wake you up and prevent you from sleeping fully.
The day before the sleep study:
- Have no alcohol. Zero.
- Stop caffeine and chocolate after noon.
- Have a couple of bananas.
The day of the sleep study:
- Have no alcohol, Zero, nada, zilch.
- Exercise as much as you should be able to do if you had no sleep issues.
- Exercise vigorously 30-45 minutes in the morning if possible, but no later than early afternoon. Get into your aerobic zone for most of it. Again, be an adult and follow the advice of your doctors and qualified exercise instructors. For me, it seems that weight bearing causes the symptoms faster than aerobic exercise, so a few squats to tire your legs out may be all you need. Climb stairs with a couple of gallons of water in each hand should do the trick as well if a gym is out of your reach.
- Make sure to cool down and stretch your legs.
- Have a couple of bananas after your workout.
In my case, it was like magic, symptoms that had never shown up before became clearly and consistently visible during the lab work. The CPAP machine that was fine for the first 6 studies was completely inadequate now.
By challenging your body, and your CPAP machine you'll have the best study possible.
If you are a sleep specialist, or researcher, and would like detailed information I would be willing to make that available privately. Please leave me a comment with information about your publicly visible contact information and I'll reach out to you.
Subscribe to:
Posts (Atom)