[Message Prev][Message
Next][Thread Prev][Thread Next][Message
Index][Thread Index]
Re: xAP Speech 1.4 Released
--------------020309060304030902060704
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
I actually gets more fiddly that just pre-loading the voices. Natural
Voices after a period of inactivity unload themselves so even after you
know the voices have been loaded you still get delay when it reloads.
The trouble with all this is that the loading/unloading and the delays
are all very vaired of different hardware so i can see alot of fiddling
to get any solution.
My only suggestion that may be worth trying would be to make a vbs
script that says 'hello' at 0 volume using all the voices you want and
trigger that script using task scheduler to run every 5 minutes. Now
this will cause a cpu blip when it runs but it should be a pretty minor
one. Hopefully it would be enough to keep the sapi voice cache always
ready. Assuming it does keep them loaded you can reduce the frequency of
the task until it stop working.
http://www.myitforum.com/articles/11/view.asp?id=9159
for the needed bits
I did make a start of adding this functionality into Speech but havn't
got that far. Certainly with the natural voices once cached you can
switch between them quickly, maybe the Neovoices are different I've
never used them
James
Paul Gale wrote:
> Ah OK - I've never heard of anything.
>
> Paul.
>
>
>
>
>> -----Original Message-----
>> From: xap_automation@xxxxxxx
>> [mailto:xap_automation@xxxxxxx] On
Behalf Of Gregg Liming
>> Sent: 12 December 2006 15:43
>> To: xap_automation@xxxxxxx
>> Subject: Re: [xap_automation] xAP Speech 1.4 Released
>>
>> Quoting Paul Gale (12/12/06 9:16 AM):
>>
>>
>>> Yes - this delay is caused by the voice loading into ram etc
and
>>>
>> initialising.
>>
>> Thanks Paul. I knew there was an init delay and should have
rephrased
>> the question to be something more like... Is there some means to
>> maintain more than one "voice" or "engine"
instance in memory (perhaps
>> optionally pre-loadable) such that the delay switching between
voices is
>> negligible?
>>
>> And, a follow-on question if the answer to the above is yes... Are
there
>> implications on using priority TTS messages to override a
non-priority
>> TTS message while also switching voices?
>>
>> Gregg
>>
>>
>>
>>
>>
>> Yahoo! Groups Links
>>
>>
>>
>>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
--------------020309060304030902060704
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01
Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1"
http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
I actually gets more fiddly that just pre-loading the voices. Natural
Voices after a period of inactivity unload themselves so even after you
know the voices have been loaded you still get delay when it reloads.
The trouble with all this is that the loading/unloading and the delays
are all very vaired of different hardware so i can see alot of fiddling
to get any solution.<br>
My only suggestion that may be worth trying would be to make a vbs
script that says 'hello' at 0 volume using all the voices you want and
trigger that script using task scheduler to run every 5 minutes. Now
this will cause a cpu blip when it runs but it should be a pretty minor
one. Hopefully it would be enough to keep the sapi voice cache always
ready. Assuming it does keep them loaded you can reduce the frequency
of the task until it stop working.<br>
<a class="moz-txt-link-freetext" href="http://www.myitforum.com/articles/11/view.asp?id=9159";>http://www.myitforum.com/articles/11/view.asp?id=9159</a>
for the needed
bits<br>
I did make a start of adding this functionality into Speech but havn't
got that far. Certainly with the natural voices once cached you can
switch between them quickly, maybe the Neovoices are different I've
never used them<br>
<br>
James<br>
<br>
<br>
Paul Gale wrote:
<blockquote
cite="mid239FFCCB86EEA04DA6757977FA6F63B40224F5@xxxxxxx"
type="cite">
<pre wrap="">Ah OK - I've never heard of anything.
Paul.
</pre>
<blockquote type="cite">
<pre wrap="">-----Original Message-----
From: <a class="moz-txt-link-abbreviated" href="mailto:xap_automation@xxxxxxx";>xap_automation@xxxxxxx</a>
[<a class="moz-txt-link-freetext" href="mailto:xap_automation@xxxxxxx";>mailto:xap_automation@xxxxxxx</a>]
On Behalf Of Gregg Liming
Sent: 12 December 2006 15:43
To: <a class="moz-txt-link-abbreviated" href="mailto:xap_automation@xxxxxxx";>xap_automation@xxxxxxx</a>
Subject: Re: [xap_automation] xAP Speech 1.4 Released
Quoting Paul Gale (12/12/06 9:16 AM):
</pre>
<blockquote type="cite">
<pre wrap="">Yes - this delay is caused by the voice
loading into ram etc and
</pre>
</blockquote>
<pre wrap="">initialising.
Thanks Paul. I knew there was an init delay and should have rephrased
the question to be something more like... Is there some means to
maintain more than one "voice" or "engine" instance in
memory (perhaps
optionally pre-loadable) such that the delay switching between voices is
negligible?
And, a follow-on question if the answer to the above is yes... Are there
implications on using priority TTS messages to override a non-priority
TTS message while also switching voices?
Gregg
xAP_Automation Main Index |
xAP_Automation Thread Index |
xAP_Automation Home |
Archives Home
|