You still need multiple microphones for directionality -- The wider you space them, the more accurate the beamforming. This is a problem we fight in hearing aids & cochlear implant speech processors due to limited size on the spine of BTE & RIC devices.
Yes, we'll definitely see SNR continue to improve and at lower price points. What is even more important, in my mind, is that a unidirectional MEMS microphone with high SNR can replace multiple-microphone arrays - today. That means, even if the per unit cost is still at a premium, the value prop still makes sense. At a high level, one device means the total system is easier to design and manufacture, less space is taken up on an already tight motherboard, and there will be fewer points of failure (not to mention lower power, heat, etc., etc.). I'm just wondering about uptake of the Harman product though because I haven't seen much written about it to know how well it works.
According to the analysts, all the high-end smartphone makers are going for as much SNR as they can get, and are willing to pay the premium price they demand. The reason is that the voice recognition and noise canceling algorithms work so much better, low-volume voice calls are easier to understand and the quality of the soundtrack on video recordings is even more important than the video quality. Of course, like everything else, as volume production ramps up and prices go down the other-than-top-line-models will start to adapt HD mics too.
In January Harman announced unidirectional analog and digital MEMS microphones with SNR (signal-to-noise ratio) up to 68 dBA. Have you come across anyone designing these in yet? If a single MEMS microphone can replace multi-mic arrays for noise cancelation, it should speed adoption by significantly reducing the cost and space required within the target device. Thoughts?
All your arguements for digital outputs are valid, and in fact the first startup's first microphone (Akustica) had digital outputs. Unfortunately, its kind of like the VHS versus Beta video take war--if you can remember that far back--everybody was already set up for analog mics (electrets) so much so that even Akustica has had to acquiesce a prodeuce analog models. But I am with you, and think hat eventually all MEMS mics will be digital.
It would be interesting to know how far digital microphones (DMICs) -- MEMS mics that include the A/D converter -- have penetrated within the overall MEMS mic market. From an interface perspective, integrating the A/D into the same package as the MEMS makes so much more sense -- especially for applications requiring many microphones (beam-forming arrays, etc.). From the perspective of the audio processor at the receiving end, the two-wire PDM data & clock interface from a DMIC is so much easier to handle than a low-level analog signal, that it's probably well worth the extra cost of the MEMS mics with built-in A/D converters.
Oh that's an answer with fact figures; my estimation was based on the observation of my own. But yes it is very true that MEMS Mics will be having more chances of acceptance in other left out application with possible use of it for betterment.