I am teaching business english in Europe. I like to play news clips
from NPR, BBC, etc. to my classes, but the speech is to fast for the
more novice students. I have tried slowing down the clips using the
pitch shift and time shift functions on my Goldwave, but the results
aren't really an improvement.

Has anyone ever heard of software that can look for the silences
between the words on speech recordings and multiply them?

I have the impression that software that does the opposite is already
in use on some of these broadcasts.

Rob

On 13 Dec 2005 07:26:46 -0800, wrote:

I am teaching business english in Europe. I like to play news clips
from NPR, BBC, etc. to my classes, but the speech is to fast for the
more novice students. I have tried slowing down the clips using the
pitch shift and time shift functions on my Goldwave, but the results
aren't really an improvement.

Has anyone ever heard of software that can look for the silences
between the words on speech recordings and multiply them?

I have the impression that software that does the opposite is already
in use on some of these broadcasts.

WaveLab does this rather neatly. The Autosplit function can be set to
detect pauses and, instead of splitting, just insert markers. Then
there's a function to insert a defined length of silence at each
marker.

wrote:
Has anyone ever heard of software that can look for the silences
between the words on speech recordings and multiply them?

Normal speech elides most of those supposed silences, which is why
voice. recognition. products. always. used. to. insist. you. talk. like.
this. ... it made determining where one word ended and another started
TREMENDOUSLY easier.

You might be able to insert some time automatically, where a pause does
actually occur, but it would still be phrases rather than words.
Interesting concept, though; sounds worth trying...

But maybe you ought to find a less frenetic broadcaster. News reporters
are always trying to speak as quickly as they can without losing
clarity. A conversational show -- an interview, for example -- is likely
to involve a more deliberately paced conversation.

(But the fact is, *MOST* native speakers are too fast for a novice
listener. Learning how to listen at that speed takes a long time. When I
was in practice, my Spanish was *almost* good enough -- I could
recognize enough phrases, and skip over words I didn't know, to get most
of what was being said.)

On Tue, 13 Dec 2005 18:28:30 -0500, Joe Kesselman
wrote:

You might be able to insert some time automatically, where a pause does
actually occur, but it would still be phrases rather than words.
Interesting concept, though; sounds worth trying...

And, of course, it's much easier to contract pauses that DO occur than
to analyse where one SHOULD be.