[castleofargh]

I'm curious if my thoughts are correct, our ears and brains can differentiate sounds in milliseconds, and it's how we are able to have spacious cues, i.e. Doppler effects or echo in a large room. So we correlate that to iems soundstage to sound decaying at some region in the fr? Can we be able to somehow interpreted from the Csd graphs?

yes small timing variations do count, no, probably not the way you think. the main spacial cues are all the perceived variations between left and right ear(direction and distance based on delays and loudness differences).
there are other cues, like frequency response to try and determine the place on a vertical axis, or how we can locate high frequencies but not so much low freqs(I guess because the period is bigger than the distance between the ears but I'm not 100% sure). and a few less important things. that can still have some part in the result.as soon as we discuss subjective notions like soundstage, you have to consider that it's a mind construct based on absolutely everything the brain could get to mix in the giant sense blender. including ideas, what you were concentrating on, how tired you are, if it's hot outside... all information, from all senses, plus ideas. the brain will discard some stuff(because else there is just too much to process), and it's not necessarily our choice to decide what is discarded. then it will try to form something with the rest that can have meaning for us. making up some missing or changing conflicting elements if it seems like it could help(the brain seems really bent on satisfying us, to a fault). so trying to reverse engineer the process and find direct and clear cause and effect, or the right cue for the right subjective result, can sometimes be like looking at poop and guessing what food it originally was. sometimes it's super simple, and sometimes, it really just looks like poop(analogy skill 3year old).


the decays as an idea of room reverb isn't a bad idea but room reverb is probably a good deal longer to decay, so I don't know if the brain would really associate both that way. still, reverb in a room is the kind of cue that the brain will try to remove(or attenuate at least). you can test it, add a lot of reverb, and rapidly you get used to it. if you turn it off and put it back on it seems like there is way more for a time. warning totally made up hypothesis follows: I imagine it might be one of those survival skills, where we remove what we slowly identify as "ambience", so that we're able to focus on what really matters and notice when the ninjas will attack.or the mosquitoes, but something deadly for sure.

 

[datranz]

What I mean is it possible that a like decay in the treble region could in there allows for a bigger sound. For example, when I compare the be compare to the dunu, the treble and vocals of the be seems to be (subjectively) juicer, more space. The be is drier, quicker and more direct, like a wall of sound. So my question is how can we mearsure or objectively describe this character.

 

[ThomasHK]

What I mean is it possible that a like decay in the treble region could in there allows for a bigger sound. For example, when I compare the be compare to the dunu, the treble and vocals of the be seems to be (subjectively) juicer, more space. The be is drier, quicker and more direct, like a wall of sound. So my question is how can we mearsure or objectively describe this character.

It's not so straightforward. At my previous company we tried to find correlation between perceived width and the CSD ringing. There wasn't any.

 

[Sinusoid]

I'm guessing it's not the mic, but the coupler that's causing the massive differences. The bass rolloffs look typical of a lack of seal to me.

Seal and size. Your coupler is MUCH too small, and you're probably leaking all around the place. Solve the leak with Silly Putty or Blue Tack; either will seal nicely and peel off cleanly without leaving a residue.

Also note that mics need to be calibrated with a given preamplifier; drive voltage will affect not just sensitivity but frequency response, as will the loading of the mic preamp. You'll get different results if you don't calibrate the entire input chain.

Thank you guys for your responses! I am very positive that the seal is perfect, the tube I use is stretch silicone and it's quite difficult to get it on. I have also confirmed the seal by very gently blowing/sucking air into/from one end when just the IEM / mic is attached and I feel a firm resistance, so no leaking. About the size of the coupler, as I have understood, any kind of "room" will have a "sound", be it a concert hall or a silicone tube, so that is why I have figured it's best to have as little space between the capsule and the tip of the speaker as possible, to minimize the "sound of room". I have tried to use a longer coupler to see the difference and I must say the bottom end does not change much, only the smoothness of the top end gets rougher, leaving me believing there are unnecessary additional reflections happening with longer tube.

I do not completely understand the comment about calibration of the mic to the preamp. I am recording studio and live sound engineer and preamp gain has little to no affect to mic response in these fields, some people sometimes deliberately overdrive the mic pre to get a certain saturated or distorted sound, but other than that, if you use the gain in the very wide proper range it really does not affect the sound any. I'm keeping about 20db of headroom to be sure there is enough headroom. I did test the same mics measuring a free field speaker and there were differences (mics do have price differences too) they did not show such large differences I am seeing with the IEM's.

So it must be something to do with the closed and close measurement. Maybe some mics suffer more from the "proximity effect"? I have access to the full range of 64 Audio, VE and Ultimate Ears in-ears and by ear, the bottom end does not sound that full on most of the models, that is why I'm interested to see the measurements, but my small experience is showing it's hard to impossible to actually have a valid measurement.

Please people, who have access to different microphones, see if you are seeing what I am - every mic giving different results!

Including a picture where I use the same coupler (tube) stretched over a tip of the custom IEM

 

[bahaja]

Sorry if i posted this in the wrong thread, but how do you interpret a sound signature from a graph?

For example, when looking at :

Having tried the MH334SR previously, i assume that the small dip on the 2.5k-4.5k makes the female vocal sounds more transparant and thin. On the other hand, a small peak on the 5-7k results in a more fleshed out string instruments and details are a little bit more improved. The plunge after 10k might explain why the 334SR still has a roll off despite its pronounced treble. Overall the iem sounds bright and neutral but it doesn't sound as coherent as the original MH334.

For the remaining two IEMS, i have not tried it and i found it difficult to understand or guess what does it sounds like. For example does the universal with a flat 1k-5k and a peak at 6k means that it has a midcentric tuning and a bit of sparkle? Or it has a midcentric tuning with a smooth and rolled off treble?

How do you guys interpret FR graphs? How does a peak or a dip at a certain frequency range affects the characteristic of a certain IEM?

Once again apologies if i posted this in the wrong thread, and using a material not original from this thread.
You can find this graph on https://twitter.com/kindo3/status/904591999075893248 and it belongs to kindo3 or Mr Keita Suyama-san of Fitear.

 

[Francisk]

Basing it off measurements made with both a 318 compliant coupler and the IMM6 with insert depth matched to resonant peaks. Here's the Carbo Tenore that I used for reference.

Why does the Tenore graph looks so overly V shaped?

 

[DanWiggins]

Thank you guys for your responses! I am very positive that the seal is perfect, the tube I use is stretch silicone and it's quite difficult to get it on. I have also confirmed the seal by very gently blowing/sucking air into/from one end when just the IEM / mic is attached and I feel a firm resistance, so no leaking. About the size of the coupler, as I have understood, any kind of "room" will have a "sound", be it a concert hall or a silicone tube, so that is why I have figured it's best to have as little space between the capsule and the tip of the speaker as possible, to minimize the "sound of room". I have tried to use a longer coupler to see the difference and I must say the bottom end does not change much, only the smoothness of the top end gets rougher, leaving me believing there are unnecessary additional reflections happening with longer tube.

I do not completely understand the comment about calibration of the mic to the preamp. I am recording studio and live sound engineer and preamp gain has little to no affect to mic response in these fields, some people sometimes deliberately overdrive the mic pre to get a certain saturated or distorted sound, but other than that, if you use the gain in the very wide proper range it really does not affect the sound any. I'm keeping about 20db of headroom to be sure there is enough headroom. I did test the same mics measuring a free field speaker and there were differences (mics do have price differences too) they did not show such large differences I am seeing with the IEM's.

So it must be something to do with the closed and close measurement. Maybe some mics suffer more from the "proximity effect"? I have access to the full range of 64 Audio, VE and Ultimate Ears in-ears and by ear, the bottom end does not sound that full on most of the models, that is why I'm interested to see the measurements, but my small experience is showing it's hard to impossible to actually have a valid measurement.

Please people, who have access to different microphones, see if you are seeing what I am - every mic giving different results!

Including a picture where I use the same coupler (tube) stretched over a tip of the custom IEM

Microphone diaphragms - especially condensers and electret-style - are very susceptible to variations in load and bias voltage. You can tilt the FR of a mic just by changing the load capacitance or the bias voltage. I've designed a lot of mics in my time (Rode, Blue, others) and it's not as straight-forward as you think. Many companies use the exact same capsule in 3-4 models and the sonics - and measured performance - is quite different due to little things like capsule loading and bias. You'll get roll-off, or peaks, at the ends of the response. It's not just a sensitivity thing.

An IEC60318-4 coupler has a volume that replicates an ear canal. You want something around 8mm average diameter, and around 25mm long. There is also a specific mechanical stiffness to the entire structure. You have essentially zero length - so add some length.

And also, realize that you will NOT get the same results as others! Unless you're running the same gear (or gear calibrated to be the same) you should expect differences. Your measurements should be consistent for your own set of data, but will probably not be consistent with others' measurement. Is that something you can calibrate out? For FR, I would say yes with quite a bit of effort. For THD and CSDs, I'm not so sure...

Lastly - EVERYONE IN THIS THREAD please remember that we are looking at sample sizes of 1 or 2! This is by NO MEANS representative of a production run! Typically you look for a 2 sigma distribution of product to determine what is "nominal", and that means at least 59 units measured. Expect to see 3-6 dB variations in products, especially in the higher end, across a production run. Yes, it really is that loose! A TIGHT run would be things like the Event Opal studio monitors where we hand-calibrate to get an industry-leading 2 dB variation max from unit to unit. EXPECT differences, look at trends, and understand that unless you are collecting dozens and dozens of measurements on dozens and dozens of samples of a given product, you have nice anecdotal evidence but it is by no means representative of a normal production run.

There's a reason we build hundreds of units for EVT and DVT testing. To not only validate the product/design, but also to get a great feel for all the tolerancing and variations that will happen during normal production and then try to tighten it up as much as possible. But without significant tuning - and that would almost have to be in the active domain - expect 3-6 dB variations as "normal".

 

[castleofargh]

what he said!!!! ^_^

 

[Francisk]

Microphone diaphragms - especially condensers and electret-style - are very susceptible to variations in load and bias voltage. You can tilt the FR of a mic just by changing the load capacitance or the bias voltage. I've designed a lot of mics in my time (Rode, Blue, others) and it's not as straight-forward as you think. Many companies use the exact same capsule in 3-4 models and the sonics - and measured performance - is quite different due to little things like capsule loading and bias. You'll get roll-off, or peaks, at the ends of the response. It's not just a sensitivity thing.

An IEC60318-4 coupler has a volume that replicates an ear canal. You want something around 8mm average diameter, and around 25mm long. There is also a specific mechanical stiffness to the entire structure. You have essentially zero length - so add some length.

And also, realize that you will NOT get the same results as others! Unless you're running the same gear (or gear calibrated to be the same) you should expect differences. Your measurements should be consistent for your own set of data, but will probably not be consistent with others' measurement. Is that something you can calibrate out? For FR, I would say yes with quite a bit of effort. For THD and CSDs, I'm not so sure...

Lastly - EVERYONE IN THIS THREAD please remember that we are looking at sample sizes of 1 or 2! This is by NO MEANS representative of a production run! Typically you look for a 2 sigma distribution of product to determine what is "nominal", and that means at least 59 units measured. Expect to see 3-6 dB variations in products, especially in the higher end, across a production run. Yes, it really is that loose! A TIGHT run would be things like the Event Opal studio monitors where we hand-calibrate to get an industry-leading 2 dB variation max from unit to unit. EXPECT differences, look at trends, and understand that unless you are collecting dozens and dozens of measurements on dozens and dozens of samples of a given product, you have nice anecdotal evidence but it is by no means representative of a normal production run.

There's a reason we build hundreds of units for EVT and DVT testing. To not only validate the product/design, but also to get a great feel for all the tolerancing and variations that will happen during normal production and then try to tighten it up as much as possible. But without significant tuning - and that would almost have to be in the active domain - expect 3-6 dB variations as "normal".

I wonder how JBL is able to achieve less than 1dB level matching for their LSR30X (Massdrop version of their popular LSR305) active speakers.

 

[crinacle]

What I mean is it possible that a like decay in the treble region could in there allows for a bigger sound. For example, when I compare the be compare to the dunu, the treble and vocals of the be seems to be (subjectively) juicer, more space. The be is drier, quicker and more direct, like a wall of sound. So my question is how can we mearsure or objectively describe this character.

I've had the NocturnaL Gorham for a while while I did my Atlantis review. It sounded extremely sharp and fatiguing and, while had the characteristic of clarity, was borderline unlistenable to me. CSD graphs show quite a significant decay in the 6k region that I've never seen before in any other IEM.

Thank you guys for your responses! I am very positive that the seal is perfect, the tube I use is stretch silicone and it's quite difficult to get it on. I have also confirmed the seal by very gently blowing/sucking air into/from one end when just the IEM / mic is attached and I feel a firm resistance, so no leaking. About the size of the coupler, as I have understood, any kind of "room" will have a "sound", be it a concert hall or a silicone tube, so that is why I have figured it's best to have as little space between the capsule and the tip of the speaker as possible, to minimize the "sound of room". I have tried to use a longer coupler to see the difference and I must say the bottom end does not change much, only the smoothness of the top end gets rougher, leaving me believing there are unnecessary additional reflections happening with longer tube.

I do not completely understand the comment about calibration of the mic to the preamp. I am recording studio and live sound engineer and preamp gain has little to no affect to mic response in these fields, some people sometimes deliberately overdrive the mic pre to get a certain saturated or distorted sound, but other than that, if you use the gain in the very wide proper range it really does not affect the sound any. I'm keeping about 20db of headroom to be sure there is enough headroom. I did test the same mics measuring a free field speaker and there were differences (mics do have price differences too) they did not show such large differences I am seeing with the IEM's.

So it must be something to do with the closed and close measurement. Maybe some mics suffer more from the "proximity effect"? I have access to the full range of 64 Audio, VE and Ultimate Ears in-ears and by ear, the bottom end does not sound that full on most of the models, that is why I'm interested to see the measurements, but my small experience is showing it's hard to impossible to actually have a valid measurement.

Please people, who have access to different microphones, see if you are seeing what I am - every mic giving different results!

Including a picture where I use the same coupler (tube) stretched over a tip of the custom IEM

Like Dan said, the coupler looks too small in diameter and in length. My own coupler is 8mm inner diameter and the tip placed roughly 15-18mm away from the microphone face.

Sorry if i posted this in the wrong thread, but how do you interpret a sound signature from a graph?

For example, when looking at :

Having tried the MH334SR previously, i assume that the small dip on the 2.5k-4.5k makes the female vocal sounds more transparant and thin. On the other hand, a small peak on the 5-7k results in a more fleshed out string instruments and details are a little bit more improved. The plunge after 10k might explain why the 334SR still has a roll off despite its pronounced treble. Overall the iem sounds bright and neutral but it doesn't sound as coherent as the original MH334.

For the remaining two IEMS, i have not tried it and i found it difficult to understand or guess what does it sounds like. For example does the universal with a flat 1k-5k and a peak at 6k means that it has a midcentric tuning and a bit of sparkle? Or it has a midcentric tuning with a smooth and rolled off treble?

How do you guys interpret FR graphs? How does a peak or a dip at a certain frequency range affects the characteristic of a certain IEM?

Once again apologies if i posted this in the wrong thread, and using a material not original from this thread.
You can find this graph on https://twitter.com/kindo3/status/904591999075893248 and it belongs to kindo3 or Mr Keita Suyama-san of Fitear.

The Titan looks bright me, especially when you normalise the graphs at 600Hz.

The 6-7K point does give the sound boosted clarity and shimmer, but that's generalising things. There's also the issue of scaling, which is a highly compressed 10dB step scale. And there seems to be quite a bit of smoothing going on. I could be wrong though.

Why does the Tenore graph looks so overly V shaped?

it does bcos it do

The Tenore is a V-shaped IEM, no doubt. But it's actually quite a tastefully done V. Nothing like the typical curves prevalent in budget chifi.

 

[DanWiggins]

I wonder how JBL is able to achieve less than 1dB level matching for their LSR30X (Massdrop version of their famous LSR305) active speakers.

Level matching isn't hard; FREQUENCY response matching is much more difficult...

 

[bahaja]

I've had the NocturnaL Gorham for a while while I did my Atlantis review. It sounded extremely sharp and fatiguing and, while had the characteristic of clarity, was borderline unlistenable to me. CSD graphs show quite a significant decay in the 6k region that I've never seen before in any other IEM.



Like Dan said, the coupler looks too small in diameter and in length. My own coupler is 8mm inner diameter and the tip placed roughly 15-18mm away from the microphone face.



The Titan looks bright me, especially when you normalise the graphs at 600Hz.

The 6-7K point does give the sound boosted clarity and shimmer, but that's generalising things. There's also the issue of scaling, which is a highly compressed 10dB step scale. And there seems to be quite a bit of smoothing going on. I could be wrong though.



it does bcos it do

The Tenore is a V-shaped IEM, no doubt. But it's actually quite a tastefully done V. Nothing like the typical curves prevalent in budget chifi.

Yeah it seems like the Titan is a bit different to Fitear's house sound. Even the 334SR with a relatively flatter response sounds a tad too bright for me.
I agree the scaling could be better, but I don't understand what you mean about smoothing. Does it mean that the data is averaged?

I found another graph from another tweet of his (Titan vs prototype) - https://twitter.com/kindo3/status/905298423615717376

The prototype looks flat and have a decent high end extension, could be a successor to the Private 435. I'm really looking forward for this product to enter production

 

[crinacle]

I agree the scaling could be better, but I don't understand what you mean about smoothing. Does it mean that the data is averaged?

It is "smoothed", yes. Makes it more aesthetically pleasing (as opposing to a jittery mess) and also eliminates large peaks and dips.

 

[ForceMajeure]

When you are JBL you have enough money to do what you want...


Of course nowadays it's all about the production line and the strategy they use to make a product at a certain price bracket.Being as a big as JBL has its advantages. The amount they spend on perfect matching and QC can be less than for others companies as they have the expertise and economy of scale.

 

[castleofargh]

I wonder how JBL is able to achieve less than 1dB level matching for their LSR30X (Massdrop version of their famous LSR305) active speakers.

it is within a given range(the graph shows 50hz to 12khz). also matching a pair does not necessarily mean buying a second pair will measure the same within 1dB. maybe it will and they just have that level of consistency? I really don't know, but that's not the claim.
headphones/IEMs matched for +/-1dB(so 2dB^_^) can show sample variations bigger than 2dB. and not all manufacturers put the same efforts into this. some will just put the first 2 drivers they grab together and move on.