Mac Forum / Applications / Mac Applications / January 2005

All:

Hmmm.... my reply was fine when I previewed it in Google.

For the record, I have never experienced this problem - but I keep
things nice and clean regarding iListen. Which is to say, I tend to
dictate into one document at a time, and commit corrections when after
I make them.

I think this problem usually occurs when people have more than one
document open in multiple applications. When this happens, iListen has
a HUGE amount of stuff in RAM. Also, keep in mind that iListen owns the
sound input channel when it is running. If someone were to allow their
computer to go to sleep with the mike open, unpredictable things are
likely to happen, since iListen wasn't allowed to finish its work
before the computer fell asleep.

There is a similar circumstance I can relate to you. Try checking 11
email accounts at once and then putting your computer to sleep while
checking is still in progress. When you wake up your computer, you will
very likely have to restart Mail so it can get its bearings again (I
have actually had this happen several times). iListen is no different.
In fact, there is a lot more going on with iListen

So while the "official word" would be to not allow your computer to go
to sleep, there are things you can do to minimize - and possibly
eliminate - the unstable behavior:

- close (or at least save) all your documents before putting the
computer to sleep
- make sure the microphone is turned off
- make sure you have committed any corrections in all open documents

Regarding "is it too much too ask for the developer to fix the
application," there is nothing to fix here. It really is an OS issue.
If iListen is put to sleep with too many threads hanging, the OS simply
can't keep track of all of them reliably and things get muddled. But if
you follow the above instructions, you should be OK.
Best Regards,

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.

All:

I think I can best sum things up this way, regarding both my on
again-off again employment with MacSpeech, and the issue of using
iListen versus not using iListen. I'll try to keep it as brief as
possible, since many of you probably know the story already, and the
rest probably don't care anyway ;).

I was first hired by MacSpeech less than a year after being laid off by
Apple where I was one of two Small Business Evangelists. I was laid off
because Steve Jobs had returned and he pruned Apple of anything that
was not focused on the most profitable areas of the company. Small
Business was not one of those areas (although I am pleased that it has
returned in a small way since then).

The company was successful in getting investment funding - a seriously
cool accomplishment for any Mac-only company. We introduced iListen 1.0
in 2000. Shortly after introducing it we ran into a problem with Mac OS
9.1. We weren't the only ones. Companies like Dantz and others had
serious problems as well. Ours could not be resolved until 9.2, so we
were significantly handicapped.

Along about this time the dot-com crash hit, and the company was hard
up for cash. We were all working for options. I had an internal
conflict with the person who was then president of the company, and we
parted ways. Things continued to get worse for the company, to the
point it could no longer afford to pay for the person who was
president. He resigned. Throughout all of this, I maintained contact
with Andy Taylor, the chief architect of iListen and the person that
started the company in the first place. He took over as CEO and hired
me back.

When 9.2 came out, things were looking up again, and Apple was even
carrying us in the retail stores. Then a new edict came out: you can't
be in the Apple stores unless you were Mac OS X compatible. The problem
was, Mac OS X Core Audio was not yet ready for what iListen had to
throw at it. IBM got around this problem by writing their own sound
drivers for the headset they were shipping with ViaVoice. We preferred
to wait until the necessary tools were in Mac OS X so we could use Core
Audio. That wouldn't be until 10.1.5.

By this time, it had been decided we would jump from version 1.2.x all
the way to 1.5. Initially this was my suggestion, based on a list of
features that were to be included in the product. As we moved forward,
however, more and more features were dropped in order to do a release
sooner rather than later. I was fine with this, but wanted to label the
version as 1.3 instead. (At the time, I was running the beta program.)

Here is where things get interesting. Although others in the company
decided to keep the version number at 1.5, it wasn't ready for
prime-time yet. Every beta release was an improvement on the last (as
you would expect), and we were getting close - but we weren't there
yet. There was serious pressure to release a product so new revenue
could be generated and we could make another attempt to get in the
Apple stores with a Mac OS X compatible product.

OK - I'll cut to the chase. There was great gnashing of teeth, and
words bandied about back and forth, but nonetheless, I woke up one day
and my email account had been shut off and another MacSpeech employee
had sent out a press release announcing 1.5. To my recollection, there
was not one beta tester who felt it was ready to be released.

I immediately sent out a very public note criticizing the release.
Anyone who knows me knows that I don't lie to achieve results, and I
demand a very high quality product from the companies I have worked for
- both as an employee and consultant. In my public announcement I did
mention that I had every confidence the company would fix the problems
in 1.5, but that there was no way I could support its release at that
time.

Within a couple of weeks MacSpeech released 1.5.1 to address some of
the bugs in 1.5, and a few weeks after that release 1.5.2, which I felt
was actually what should have been called version 1.3 - and was the
first Mac OS X version of the software I consider usable.

MacSpeech continued to release new versions - as I knew they would -
and 1.6 had many of the bells and whistles that were supposed to be in
the product when I first suggested a jump to version 1.5. About a year
after being thrown out of the company because I wouldn't support the
release of buggy software I was asked to return. I did so after serious
discussions with Andy regarding the nature of how things would be
released in the future.

To sum up, MacSpeech is a very small company that is something of an
anomaly: it continues to survive and grow with absolutely no outside
funding at this point. Every advance we make is made because of the
work of those who believe in what we are trying to do. Even while I was
not working for the company, Andy and I remained friends. I believe in
the company, its employees, and its end users. All make a good team.

For those of you who would decide to not support us because we don't
place nice with Apple's sleep (regardless of whose fault it is), I ask
that you be patient and allow us to work though these issues.

Changing speech to text is an incredibly difficult task, and you will
get much farther by supporting us than ditching us, since we are the
only ones willing to tackle the problem on the Macintosh. And MacSpeech
has stood by some very ambitious goals - such as the ability to not
only dictate into any application, but also employ correction while
within those applications as well. Something no other application does
on either platform.

Bottom line: when I thought the company had veered too far away from my
core "scruples" (for lack of a better word), we parted company. I would
not be working for them now if I thought they put out a product that
sucked, or if I found out the intended to do so.
Best Regards,

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.