[DECtalk] GitHub DECtalk is really coming along quite nicely

[Don]

On 10/16/2022 1:24 PM, Karen Lewellen wrote:
> Wait, why buy anything if you do not understand, and research enough to act
> upon, the financial benefits for the company?

People ROUTINELY make bad decisions -- business or otherwise (do you sit
down and analyze the financial implications of buying a particular car or
home?  do you know what the total cost of ownership will be?  do you
research the costs of alternatives and make a black-and-white decision
based solely on those numbers?)

<https://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect>

The decision to purchase a company (or a technology) is often made
by "money people".  They look at the costs, assets, liabilities and
opportunities and HOPE to make a decision that yields a profit.

You wouldn't buy a house without ensuring that the title is "clean",
inspecting the property to determine its condition, checking out the
neighborhood with an eye towards how current trends might affect
the future value of the house, etc.

Likewise for a company -- you'd want to "have a look at the books".

But, you can't just look at a technology and understand its value
*or* it's liabilities.  Especially if you are a "money person" with
no real knowledge of the technology or implementation methodology
involved.

When you look at a piece of code, how do you put a number on the
"liabilities" (bugs) that are associated with the codebase?  You
don't even know how MANY there are, how serious they are or what
the costs of repairing them is likely to be!

And, while you are "fixing" these things (just to keep your place
in the market for THAT product), you aren't able to work on a
NEW product or explore new technologies.  Until you find that
YOUR product sees so little demand that you are now playing catch-up.

And, if you are a "tech person", you likely don't understand the business
aspects of owning that technology.  And, likely OVERESTIMATE (see above link)
your ability to maintain and enhance it.  As well as overestimating your
ability to DO the work that you think you do!

For example, a typical programmer writes about a *dozen* (or two) lines of
code per day.  I.e., if this email was a program, it would represent a few
WEEKS of work!

How could that be?  Surely I can TYPE much more than that!  Yes, but you
have to debug it, document it, attend staff meetings, cope with design
changes, interface with other workers, training, etc.

I can easily get an estimate of the size of a piece of code.  But, getting
an estimate of its complexity is a different thing, entirely.  And, trying
to estimate how long it will take me to MASTER the technology even longer
still!

> after all, these days the need to provide quality speech goes far beyond what
> gets discussed here.

And *other* technologies are used to produce that quality speech.  They
target different markets with (cumulatively) deeper pockets.

> and, given how many resources can be leveraged, size of company is  quite
> unimportant.

Leveraging anything means putting money on one side of the fulcrum.  If
you don't have the funds, then you can't apply the leverage.

I rely on the good will of many of my colleagues to critique and
exercise/debug my designs.  Work that I won't have to do.  But, without
that goodwill (paid for, in advance, by MY similarly extended good will
for their needs), I'd have to pay people to perform those tasks.

If I want to record a voice for diphone synthesis, I need to pay a
voice actor.  And, rent studio time.  And, post-process the recordings.

Repeat for the next voice.

How many such voices do you think I am going to create?

> making it seem nonsensical for this company to have eloquence at all, without
> knowing how to make it profitable.

Why would anyone buy the rights to DECtalk?  It's been obsolescent since
it was originally created -- anyone in the industry would be able to
predict that processors would get faster and cheaper and "more expensive"
algorithms would quickly displace "more efficient" ones.

Yet, it kept getting picked up by successive owners each time the
current owner came to the realization that it was NO LONGER PROFITABLE
to continue owning!  (so, why WILL it be profitable for the new owner?)

My (hardware) DECtalk units are big and expensive to produce.  The
DTC-01 has two processors -- one to handle the letter-to-sound
rules and communication interfaces and the other (a DSP) to generate
waveforms.  The labor alone must have been $100 or more!

Yet, my DECtalk workalikes fit *in* an earbud -- along with the bluetooth
radio -- and can be produced for a few dollars.  Give me a bigger box and
a bigger "cost" budget (which wouldn't be prohibitive in many applications)
and I can make considerably *better* speech!