Best quality speech synthesis

I have an application for the best possible, most natural speech synthesis
or text to speech application. This is for the creation of voice prompts for
a computerised self service system. It doesn't really matter how much
computing power it takes or how long to create each prompt since this does
not have to happen on-line or in real time. We would need very good
facilities for adjusting the emphasis, volume, pitch and timing of
individial words and syllables. Some manual tweaking of each phrase (by
someone who is not a speech expert) is acceptable.

Currently we use a professional speaker in a studio to record these prompts,
but this is far from ideal, since it's expensive, time consuming, and if we
need to add new prompts it's very difficult to match the sound of the
existing recording, especially if the speaker we used initially is no longer
available.

I've evaluated a couple of packages and services for this, but they don't
give the quality or naturalness that we're looking for. They were both
designed for automated telephony, which obviously places severe limits on
the amount of processing required to generate the speech since it has to
happen in real time. I don't have this constraint, so I'm looking for
something that does a better job, but probably takes more processing.

Do you know of such a system?

Do you know of an expert in this area, or anyone researching in this field
that I could ask?

Many thanks - Rowan

On Apr 10, 8:39 pm, "Rowan Sylvester-Bradley" <ro...@sylvester-
bradley.org> wrote:
> I have an application for the best possible, most natural speech synthesis
> or text to speech application. This is for the creation of voice prompts
for
> a computerised self service system. It doesn't really matter how much
> computing power it takes or how long to create each prompt since this does
> not have to happen on-line or in real time. We would need very good
> facilities for adjusting the emphasis, volume, pitch and timing of
> individial words and syllables. Some manual tweaking of each phrase (by
> someone who is not a speech expert) is acceptable.
>
> Currently we use a professional speaker in a studio to record these
prompts,
> but this is far from ideal, since it's expensive, time consuming, and if
we
> need to add new prompts it's very difficult to match the sound of the
> existing recording, especially if the speaker we used initially is no
longer
> available.
>
> I've evaluated a couple of packages and services for this, but they don't
> give the quality or naturalness that we're looking for. They were both
> designed for automated telephony, which obviously places severe limits on
> the amount of processing required to generate the speech since it has to
> happen in real time. I don't have this constraint, so I'm looking for
> something that does a better job, but probably takes more processing.
>
> Do you know of such a system?
>
> Do you know of an expert in this area, or anyone researching in this field
> that I could ask?
>
> Many thanks - Rowan
>
> ** Posted fromhttp://www.teranews.com**

This is about as good as it gets....

http://www.cereproc.com/press061030.html