Audio rate change -- function of user audiogram

At 2001-01-25 08:42, David Poehlman wrote:
>I have a question and a comment:
>question: is there an example of a "stylistic effect"?
>comment.  With pitch supression as it is done through speech synthesis
>today, the audio can be much slower than 80 percent and still be
>understood.  I can get my accent to comprehend quite low rates for
>instance at 50 percent.  if the voice is pre-recorded and synched then
>of course, at low rates much like a recording of any kind, it falls
>off the low end of the comprehensibility scale.  A further note.  many
>animations are accompanied by synthesized speech such as in ms agent.
>The characters move and speak but we currently have no control over
>the motion or the audio.

One customization that I would like: Allow my audiogram (hearing loss
vs frequency) to shape the algorithm used for rate change without pitch 
shift. This is significant to me, as I have 60db loss between 2Khz and 
3Khz. I would like to have the plosives to be stretched, to reduce their 
high-frequency harmonics. Does anyone have any experience with this 
possibility?
An analogous technique might be important to shorten low-frequency phonemes
(such as changing to a female voice to avoid the too-low fundamental).
My customization desire would apply to listening to narrations in SMIL,
as we need for digital talking books.

Regards/Harvey
>...

Received on Thursday, 25 January 2001 14:21:25 UTC