That is not remotely what has been said. It was an integration...

That is not remotely what has been said.

It was an integration issue. Inadequate voltage was being delivered from the ASIC to drive some or all of the elements.
A mod to allow this to be adjusted will be implemented on the MEMS wafer. This mod is claimed to be relatively straightforward. My guess would be that the mod is to the circuitry that carries the switching signal to the elements rather than the elements themselves.

They have not said they will have to make more. They are advancing (some) unfinished wafers (with the MEMS devices on them) for further processing, but this is after incorporating a proposed modification that they expect will fix the shortcoming. This possibility has been signalled to SH several times.

Do not ignore the fact that the MEMS has already been successfully driven and characterised by AP's purpose designed testing facility.

From the 20/4/16 release we read:

"The tests conducted on Phase IV MEMS structures revealed the chip to be perfectly functional when the testing reached the 1-trillion cycle milestone. Both acoustic loss and electromechanical integrity remained stable over the entire test span. “This interim milestone would be equivalent to having the speaker chip play continuously for over 10 years” said Yuval Cohen, the company’s Chief Technology Officer.
Driving the rationale for change in audio speakers is the ever-increasing demand for smaller, thinner, clearer sounding, more power-efficient speakers. Audio Pixels’ patented technologies employ entirely new techniques to generate sound waves directly from a digital audio stream using low cost micro-electromechanical structures (MEMS), rather than conventional loudspeaker elements. The company believes this innovation will enable the production of speaker products that deliver performance that is many orders of magnitude better than conventional speaker technologies all in a semiconductor chip form factor.
The tested structures are part of the company’s fabrication verification program implemented to mitigate risks in the fourth and final phase of the company’s commercialization development plan. The verification program is designed to enable the company’s development teams to examine, test, and validate production wafers for progress and compliance of the MEMS chips during and throughout the fabrication process.
“This milestone is significant as its confirms to the extent possible, that the design and fabrication refinements undertaken as a result of Phase III, deliver the performance and the reliability sought for our mass production product” added Shay Kaplan, the company’s Chief Scientist."

From the 31/5/16 CEO presentation Danny Lewin stated:

"...Additionally we implemented a risk mitigating approach to the fourth phase - for example we decided to independently fabricate 8 separate batches of wafers, and each batch of wafers holds a number of wafers in reserve before undergoing critical fabrications steps as to allow for the possibility of taking corrective measures should it be deemed necessary..."

This follows from earlier statements by AP's designers, such as the 28/5/15 presentation:

"...As has been reported, the third phase entailed receipt of completed MEMS structures with all its functional layers. The structures were meticulously and exhaustively tested and measured. We took tens of millions (of) micro scale measurements, many of which demand synchronized precision measured in nanoseconds..."

This is a delay, a disappointing delay for sure, but nothing like you have described.