Possible bug with sd Ibmtts backend

[jon orcauser]

Hi,

using ibmtts via sd 0.6.7:

in the /etc/speech-dispatcher/modules/ibmtts.conf

ibmttsUseSSML 0

spd-say '<' # says less than
the log says:
Sun Mar  1 17:48:36 2009 [152744]: Ibmtts: Type: 0, bytes: 19,
requested data: |<speak>&lt;</speak>|
Sun Mar  1 17:48:36 2009 [153308]: In stripping ssml: |<|

when we do
>>> import speechd
>>> c = speechd.SSIPClient('test')
>>> c.char('<')
>>> c.char('>')

we see the following:

Sun Mar  1 17:51:38 2009 [331358]: Ibmtts: Type: 2, bytes: 1,
requested data: |<|
Sun Mar  1 17:51:38 2009 [331455]: In stripping ssml: ||
---
Sun Mar  1 17:51:40 2009 [723844]: Ibmtts: Type: 2, bytes: 1,
requested data: |>|
Sun Mar  1 17:51:40 2009 [723908]: In stripping ssml: ||

Should the SSIPClient.char() really call the stripping ssml function?
And if so, surely we should be stripping full SSML tags, rather than
what might look like one. Note that setting ibmttsUseSSML to 1 gives
correct output again.

when using espeak:

spd-say '<' # We hear r angle r angle stroke speak l angle
Sun Mar  1 19:17:53 2009 [818982]: Espeak: Requested data:
|<speak>&lt;</speak>| 0 19

spd-say '>' # we hear l angle
Sun Mar  1 20:59:19 2009 [256805]: Espeak: Requested data:
|<speak>&gt;</speak>| 0 19

if we use the SSIPClient.char Everything seems to be fine.

I dont know why we are speaking '<' diffrently from '>' the markup
shown seems to be right. I set the debug level to 5 in the ibmtts.conf
and espeak.conf.

Thank you.

-Jon