Skip to main content

TTS RA (Text To Speech)

2 replies [Last post]
abhayani
Offline
Joined: 2005-04-04

Hi Guys,

For long we have been thinking on having TTS functionality for mobicents and here is the initial thought http://groups.google.com/group/mobicents-public/web/mobicents-tts-ra

Let us share the ideas here to give the final shape to TTS RA.

initial idea is such that TTS RA need not be asynchronous as the service using TTS RA would require the output immediately. I don't see any point in using Events for TTS RA. Service's making use of *only* TTS RA doesn't make any sense and has to be used in conjunction with Media RA to transmit the speech to SIP Client.

Questions that comes to my mind are
1) Is it required to have TTS as asynchronous?

2) Should we make TTS capable enough to accept the SDP and transmit the speech back to SIP Client? Wouldn't this be repetition of work since Media RA is already capable of doing this?

3) TTS RA makes use of FreeTTS which is not 100% compliant to JSAPI. JSAPI is not mature enough to give output as audioFile. Do you think there is any need to keep it JSAPI compliant?

4) Should TTS RA be event driven?

amit.bhayani

Reply viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
ivelin
Offline
Joined: 2003-07-13

Nice initiative, Amit.

> Hi Guys,
>
> For long we have been thinking on having TTS
> functionality for mobicents and here is the initial
> thought
> http://groups.google.com/group/mobicents-public/web/mo
> bicents-tts-ra
> Let us share the ideas here to give the final shape
> to TTS RA.
>
> initial idea is such that TTS RA need not be
> asynchronous as the service using TTS RA would
> require the output immediately. I don't see any point
> in using Events for TTS RA. Service's making use of
> *only* TTS RA doesn't make any sense and has to be
> used in conjunction with Media RA to transmit the
> speech to SIP Client.
>
> Questions that comes to my mind are
> 1) Is it required to have TTS as asynchronous?

depends on how long it takes to convert text to speech.
If you are feeding 2 words, it probably makes no sense to use asynchronous interface.
However if you are feeding a page of text or an input stream which can feed additional text over time, then you don't want to block.

Maybe consider alternative methods. One which takes a short string and returns the url of the resulting file immediately. Other methods can take as input text, inputstream or url of a file and fire events on a TTS session as progress is being made or errors occur.

>
> 2) Should we make TTS capable enough to accept the
> SDP and transmit the speech back to SIP Client?

Not initially. Maybe in a future version depending on user demand.

> Wouldn't this be repetition of work since Media RA is
> already capable of doing this?

right. let's postpone.

>
> 3) TTS RA makes use of FreeTTS which is not 100%
> compliant to JSAPI. JSAPI is not mature enough to
> give output as audioFile. Do you think there is any
> need to keep it JSAPI compliant?

no apparent reason.

>
> 4) Should TTS RA be event driven?

not sure what you mean

Ivelin

abhayani
Offline
Joined: 2005-04-04

1)
>Maybe consider alternative methods. One which takes a short string and returns the url of the >resulting file immediately. Other methods can take as input text, inputstream or url of a file and fire >events on a TTS session as progress is being made or errors occur.

Ok this makes sense for larger files. I will try to come up with Events type and description for this.

4)
>not sure what you mean
Like said in point 1) for larger file TTS RA should do it asynchronously and hence make use of Events to notify the services.