Prev: Accessibility note - iTunes plugins cause VoiceOver crashes
Next: iPad - Video import
From: James Jolley on 20 Jun 2010 07:29
On 2010-06-20 12:17:11 +0100, me9(a)privacy.net (Bella Jones) said:

> I put OT because I would use it on my PC if a PC one was good enough and
> better than Mac.
>
> Basically, I was wondering how good VR has become these days.

It has certainly improved, though it's not ideal in the cases your
describing - see below.
>
> Is it good enough to reproduce several different voices, from a meeting,
> for example, without knowing them first?

I'd say a most probably not. The idea of the VR systems is that they
maintain a profile of a speaker, learning intonation and such as it
goes. What you're interested in having is a transcription system i'm
guessing. I've often considered that VR research is fixed in to this
one size fits all methodology. I have this grand idea that VR could use
the cloud to maintain profiles and use mass data to compare the
intonation serverside. If not that, the idea of needing a profile to me
seems rather pathetic, given that people's voices do change over time.
>
> Some of my bread and butter work is sadly giving me RSI (I need to ditch
> it anyway) and someone suggested I look at VR again.

If you are using it within acceptable parameters, just speaking in text
yourself and having the system transcribe it, you're fine and the
accuracy is certainly good enough for an individual.

Mac speech dictate for mac also uses the dragon naturally speaking
engine, so if you don't want to move away from mac, it's worth trying
that.

Best

-James-

From: Bella Jones on 20 Jun 2010 08:54
James Jolley <jrjolley(a)me.com> wrote:

> On 2010-06-20 12:17:11 +0100, me9(a)privacy.net (Bella Jones) said:
>
> > I put OT because I would use it on my PC if a PC one was good enough and
> > better than Mac.
> >
> > Basically, I was wondering how good VR has become these days.
>
> It has certainly improved, though it's not ideal in the cases your
> describing - see below.
> >
> > Is it good enough to reproduce several different voices, from a meeting,
> > for example, without knowing them first?
>
> I'd say a most probably not. The idea of the VR systems is that they
> maintain a profile of a speaker, learning intonation and such as it
> goes. What you're interested in having is a transcription system i'm
> guessing. I've often considered that VR research is fixed in to this
> one size fits all methodology. I have this grand idea that VR could use
> the cloud to maintain profiles and use mass data to compare the
> intonation serverside. If not that, the idea of needing a profile to me
> seems rather pathetic, given that people's voices do change over time.

Yes, a transcription system is what I'm after. Thought we might not be
there yet.

> > Some of my bread and butter work is sadly giving me RSI (I need to ditch
> > it anyway) and someone suggested I look at VR again.
>
> If you are using it within acceptable parameters, just speaking in text
> yourself and having the system transcribe it, you're fine and the
> accuracy is certainly good enough for an individual.
>
> Mac speech dictate for mac also uses the dragon naturally speaking
> engine, so if you don't want to move away from mac, it's worth trying
> that.

Thanks!


--
bellajonez at yahoo dot co dot uk
From: Vivienne Dunstan on 20 Jun 2010 13:15
Bella Jones <me9(a)privacy.net> wrote:

> Is it good enough to reproduce several different voices, from a meeting,
> for example, without knowing them first?

I don't think so. I use MacSpeech Dictate from time to time, and it
needs to be trained to the voice before use. And it can only transcribe
one voice at a time. Even then there can be lots of transcription
errors. I don't think it would have a hope of coping with lots of
untrained voices. And the same principles apply to Dragon on Windows.

Viv
 | 
Pages: 1
Prev: Accessibility note - iTunes plugins cause VoiceOver crashes
Next: iPad - Video import