Re: Running Windows on Mac

Mark Conrad <none-of@xxxxxxxxxxxxxxxxxxxxx> wrote:

In article
<1j31mew.16aefy71utsqhjN%real-address-in-sig@xxxxxxxxxxxxxxx>, Rowland
McDonnell <real-address-in-sig@xxxxxxxxxxxxxxx> wrote:

So, without Windows you wouldn't be able to get the data off the
recorder in the first place?

Mark Conrad wrote:
That is correct.

Eh? I've got a digital audio recorder here and it can feed data into
anything that understands USB disc drives; and the file format is WMA or
WAV, both of which are understood by my Mac - and can even be edited if
you've paid to unlock QT Player's full set of abilities.

- - - and - - -

Again, we are forced to use the Vista partition of
the Mac. Bummer.

How come you can't use it on the Mac side?

Has to do with the MacSpeech speech app itself, which only recognizes
its own proprietory internal format, not WAV or WMA or any other
commonly used audio format.

Bastards. I can understand not working with lossily compressed formats,
but a refusal to work with something like WAV or AIFF is just wilful
bastardosity.

MacSpeech claims they will eventually support digital recorders,
but as of right now they do not.

Pfft. It's not like it'd be at all hard to do so.

Same situation on the Windows Vista side, where the "Dragon" speech
app only recognizes _its_ proprietary internal format, which is
identified as audio files with the .dra suffix.

That's just the behaviour of total bastards, that is. Lock it all in to
*OUR* kit so no-one can even use a different mode of sound input than
the one which uses our kit... Sortathing.

Difference is, there is built-in software within Dragon itself that
allows it to further process the .dra files and convert them to
text.

Are you suggesting that MacSpeech *cannot* convert sound file containing
speech into text? It seems like it, but that makes no sense at all.

Some people have tried feeding the audio directly from the speaker
of the digital recorder to a regular MacSpeech microphone,
however the resulting text has too many mistakes to be usable.

Well, yeah. Can't the electrical signal from a recorder be squirted
directly into the socket that the MacSpeech mic plugs into? Or are
MacSpeech mics digitally connected to a host computer?

In that case, we wear a wireless microphone, a brand
named "Airline 77", which has a range of 300 feet.

(NOT Bluetooth!)

Ah - so it's one of the illegal ones, is it?

No, it is a regular 800 MHz RF model built by Samson.

From what I've read, that's - umm. STFW. Now I'm confused. I dunno.

I'd previously heard that most such mic setups weren't legal. I've just
tried to find out the facts. And you know what? I failed...

It seems that electromagnetic spectrum allocation in the USA is not done
as it is elsewhere.

Thing is, Bluetooth system has too low a sampling rate,
sampling the instantaneous analog audio in the process of
turning that audio into a digital ones and zeros signal,
higher sampling rates are required by speech apps for
best accuracy.

Nothing better than 40-odd kHz can be needed for audio signals intended
for humans - and for typical speech work, 6kHz sampling rate is adequate
(that's equivalent to normal telephone bandwidth). What sampling rate
do Bluetooth rigs use?

(higher than 40kHz sampling rates can be useful because they make it
easier for the electronics to turn back into analogue and for that
matter give the electronics and easier time at the A-D end, but you've
got all the information you need from the human point of view if you're
sampling at something over 40kHz. It's why 44.1kHz was chosen as the CD
sampling rate. More bits per sample is always handy - you can't have
too many bits per sample. But excessive sampling rates are pointless
AFAIK - no-one records up to 50kHz if they want to record sound for
humans. Why bother?

....except that they do both with `super audio CD' for some reason. It's
got a sampling rate (apparently) of 2822.4 kHz, but that's the quoted
rate which is for a sigma-delta (sum-difference, 1 bit) D-A output and a
different sort of sampling rate to the one I'm talking about above.

The quoted frequency response is up to 50KHz, so SACD has an equivalent
sampling rate of about 100kHz - there is no use in providing that
bandwidth that I can think of.

Airline 77 runs higher sampling rates
than Bluetooth.

Here in the USA we have finally converted to digital TV,

Started to, surely? The conversion process ain't finished until the
analogue signals are turned off.

so the darn Samson Airline 77 is way too close to the
digital TV frequencies. (800 MHz range)

You're not worried about interfering with aviation, then? 'cos bits of
the 800s MHz are so used in the USA. The US spectrum allocatoin chart
specifies the 800s MHz range as being `land moble/fixed' or `land
mobile' or `aeronautical mobile' in adjacent bands, but I don't know
what those term actually /mean/.

Whatever you do, avoid the present Airline 77 model,
as a newer model in the 642 to 646 MHz range will be
out very shortly, in a matter of weeks.

<puzzled> Some USians seem to think that that part of the spectrum is
definitely illegal for use by such gadgets. It's in the `TV channels
21-36 range,' is that.

<http://www.ntia.doc.gov/osmhome/allochrt.pdf> says that 512MHz to 698
MHz is for telly (aside from a sliver in the middle that the radio
astronomers need to keep clear, but the US authorities permit even that
to be used by `land mobile' whatever that is - 608-614MHz).

Hmm - we seem to have something similar in the UK.

<http://www.ofcom.org.uk/radiocomms/isu/ukfat/>

`Screw the astronomers' seems to be the idea...

But what's `mobile' mean in these tables? I dunno - that could be
officialese `yer okay for radio mics'.

I mean, it's all well and good publishing such information, but not so
good when they don't seem to provide a key to what the information
means.

Noise rejection is great, the pounding racket from
the treadmills does not bother it at all.

How are you using the mic, exactly? I'm curious as to this
report. I'm thinking it's probably down to the particular
directional characteristics of the mic in question.

That and several other factors, such as the mic is extremely
close to the lips, so your voice tends to overwhelm any noise.

That's not the microphone rejecting noise, that's you improving the
signal to noise ratio in the first place.

Mic is specifically designed for a sweaty TV exercise instructor,
bouncing up and down while conducting his TV exercise or yoga or
aerobic dance routines with his "students".

This is the Airline 77 you're talking about?

Don't ask me why they want a text record of the sessions, I
suppose it is because a yoga dance student might break her leg
while she is attempting to wrap it around her neck.

Umm. You have met the concept of lesson plans? They apply to any
lesson. Maybe the idea is to review what the instructor actually did
and compare it to what was meant to be done? I expect you've got a
point - I'd want to cover my back with a recording in case of someone
involving one of those lean and hungry lawyers you have in the USA...

That setup allows an average of 99% recognition
accuracy, much better than the digital recorder
approach, which is only 98%.

This is curious.

Yes, especially considering that the Airline 77 was not
even designed for speech recognition usage.

Umm. Erm. You don't need a mic designed for that job. You need a good
mic that's got good directional characteristics so as not to pick up
much noise, used in a fashion so as to collect an accurate analogue of
the original signal with little noise.

However, according to the resident speech experts in
the knowbrainer.com forums, the Airline-77 is one
of the best mics when it comes to both accuracy and
noise rejection, even when compared to wired mics
like the very expensive Sennheiser mics.

Quite - the point is that a good mic is a good mic. Accuracy? That's
not a measurable parameter. As for noise rejection - well, what sort of
noise? What do you mean by `noise rejection'?

Look, you don't get noise rejection from a mic; that's not what goes on.
What you can and do get is `a better signal to noise ratio than you
might have thought you would get' due to the way in which it's used
(close to the mouth) and its directionality and so on.

I've got a dictaphone with switchable noise rejection, so to speak.
It's got switchable degrees of directionality (there's a stereo mic and
fancy electronics behind it) and selectably different thresholds for the
automatic gain control. This adds up to very different apparent noise
rejection - or different SNRs in different situations as I judge things.

Any given rig will be more suitable for some jobs than others, and
suitability will be apparently different depending on how the things are
used.

You get different sorts of `noise rejection'. The only one that I think
of of as actual noise rejection is the rejection of noise injected into
the mic assembly by mechanical impulse; and the `rejection' of noise
from the air by way of the microphone's inherent directional
characteristics (it's more `not detecting many dBs of noise compared to
signal' rather than `hearing the noise and ignoring it' which is what I
call `noise rejection').

What's good for a given application depends on a lot of factors - but
you've got to be doing something really odd to want a mic that doesn't
do its level best to wipe out all the noise injected into the mic
assembly by direction mechanical impulse.

BTW, although the MacSpeech voice recognition is
buggy as all heck, it still is the only halfways decent
speech app available for the Mac, assuming one buys
the most recent version 1.5 or better, and avoids
other earlier versions like 1.3

For example, yesterday I taught MacSpeech to
recognize the medical word:

"mercaptoundecahydrododecaborate"

(took me 3 minutes to train)

...as in sodium mercaptoundecahydrododecaborate,
an exotic man-made compound used in conjunction
with slow neutron bombardment radiation,
to kill cancerous tissue.

Just out of interest, when doing that sort of thing, what's a typical
neutron source? Off the top of my head, I can think of nuclear reactors
and (for smaller scale work) cold fusion rigs as sources of neutrons
used by researchers.

To make certain it recognized the word I spoke
it into MacSpeech ten times.

Recognized it perfectly all ten times, but I think
I broke my tongue in the process of speaking.

Blimey. I've never met speech recognition that worked remotely as well
when I've tried training it.

Rowland.

--
Remove the animal for email address: rowland.mcdonnell@xxxxxxxxxxxxxxx
Sorry - the spam got to me
http://www.mag-uk.org http://www.bmf.co.uk
UK biker? Join MAG and the BMF and stop the Eurocrats banning biking
.

Loading